Muscle-powered Soft Robotic Ventricular Assist Devices

Submitted in partial fulfillment of the requirements for the

degree of

Doctor of Philosophy in

Biomedical Engineering

Jooli Han

B.E., Biomedical Engineering, State University of New York M.S., Biomedical Engineering, State University of New York

Carnegie Mellon University Pittsburgh, PA

November 2020

© Jooli Han, 2020

All Rights Reserved Acknowledgements

I am thankful for everyone that took parts in this meaningful journey towards a doctorate degree. I would like to thank my doctoral advisor, Professor Dennis R. Trumble, Ph.D., for his guidance in academics and in life throughout my time here at Carnegie Mellon University. He pushed me forward by living an example and sharing bits of life wisdom. It has been a privilege to have a caring, warm-hearted mentor who supervises his lab with great respect. Thanks to his unfailing patience and support, I was able to maintain a high standard of excellence and forge my own path independently.

I would like to thank my thesis committee members for ensuring my research in the right direction. Professor Keith E. Cook, Ph.D. and Professor Conrad M. Zapanta, Ph.D. from

Biomedical Engineering Department, Professor Carmel Majidi, Ph.D. from Mechanical

Engineering Department, and Dr. Michael Scott Halbreiner, M.D. from Allegheny General

Hospital generously shared their time and expertise. Without their critical questions and opinions, this thesis would not have been completed. I cannot express how lucky I am to have been a part of the Biomedical Engineering Department of Carnegie Mellon University. Thank you, faculty and staff, especially Maryia, Keri, Misti, and Kristin, for making the friendly and supportive environment to live in every day.

This work would not have been possible without collaborations with Flexial Corporation

– thank you, Greg Peters – and Professor Sung Hoon Kang, Ph.D.’s lab from Mechanical

Engineering Department at Johns Hopkins University – thank you, Professor Kang and Ozan Erol,

Ph.D. I would also like to thank National Institutes of Health for funding this project (NIH R01

37124.1.1090440).

Thank you, Professor Wei Yin, Ph.D. and Professor David A. Rubenstein, Ph.D. from

Biomedical Engineering Department at Stony Brook University for shaping me into a young scientist. Thank you, Professor Boyle Cheng, Ph.D., for being the best professor to TA for. Thank

iii you for giving me opportunities to lecture, guide, and interact with students as a graduate teaching assistant. Thank you, Mrs. Powell, Mrs. Ganus, and Coach Cordero from Arlington High School.

I was fortunate to receive the Innovation Fellowship (180145.620.284.100000.01) supported by Tepper School of Business at Carnegie Mellon University. But more importantly, I was privileged to have met and been mentored by great people as an Innovation Fellow. Thank you, Reed McManigle, J.D. from CMU Center for Technology Transfer and Enterprise Creation, for your entrepreneurial mentorship and legal and regulatory guidance. Thank you, Seth

Boudreaux, Ph.D., for legal consultation and patent filing. Many thanks to EIRs including Lynne

E. Porter, M.D., Melanie Simko, Kit Needham, and MaryDel Brady.

I am deeply grateful for my friends and community that supported me through the toughest times. Special thanks to the people without whom this journey utterly would not have been possible.

Thank you, my lab mates – Elaine Soohoo, Ph.D., Edgar Aranda-Michel, and Matthew Kubala – for your insightful comments and hard work. My office and lab neighbors made coming into work truly joyful every day. Thank you, Angela Lai, Ph.D., Rei Ukita, Ph.D., Suji Shin, Erica Comber,

Deepshikha Acharya, Alex Rüsch, Ph.D., Sahil Rastogi, Ph.D., Raghav Garg, and more BME peeps. I would also like to thank my Korean friends for the precious memories in Pittsburgh. Thank you, Hyokyung Kim, Youngjoo Son, Haewon Jeong, Ph.D., Yunsik Ohm, Wooshik Kim, Hyung

Woo Kim, Kiwan Maeng, Jungeun Kim, Sungho Kim, Ph.D., Suyoun Kim, Ph.D., Min Kyung

Lee, Ph.D., Junsung Kim, Ph.D., Seungwhan Moon, Ph.D., SunJeung Yoon, Hyegyeong Park,

Ph.D., Minji Yoon, Daye Nam, and many more. Thank you, Phil Phelps for the awesome CBT.

Finally, I would like to express my greatest gratitude to my family. Thank you for being the best sister, Gyuri Han. Thank you, mom (Hyunjoo Lee) and dad (Youngsik Han), for your boundless love and endless support. 감사하고 사랑합니다.

iv

Abstract

Congestive heart failure (CHF) remains one of the most costly diseases in the industrialized world, both in terms of healthcare dollars and the loss of human life. This epidemic is responsible for over $40 billion dollars per year in medical costs and lost productivity, and worse, 280,000 deaths each year in the U.S. alone. Despite great strides made in the treatment of CHF using mechanical ventricular assist devices, more than half of those who develop CHF die within 5 years of diagnosis. This is because conventional ventricular assist devices (VADs) continue to be extremely problematic with long term use due to infections caused by percutaneous drivelines and the persistent risk of clot formation associated with blood-contacting surfaces.

To address both these longstanding problems, we have developed two types of completely implantable, non-blood-contacting circulatory support systems in this thesis work. An implantable muscle energy converter (MEC) was previously developed in this lab and operates by converting the contractile energy of the latissimus dorsi muscle (LDM) into hydraulic power that can be used to drive any pulsatile blood pump with power requirements consistent with steady-state

MEC/LDM output capacity. The two main advantages of this implantable power source are that it significantly reduces infection risk by avoiding a constant skin wound created by percutaneous drivelines and improves patient quality-of-life by eliminating all external hardware components.

In this thesis, we combined this unique biomechanical power source with 1) an extra-aortic balloon pump (EABP) to make a muscle-powered extra-aortic counterpulsation VAD (eVAD) and 2) a soft robotic direct cardiac compression sleeve (DCCS) to make a muscle-powered cardiac compression copulsation VAD (cVAD).

The eVAD compresses the external surface of the ascending aorta during the diastolic phase of the cardiac cycle, offering increased cardiac output and improved coronary perfusion

v without touching the blood. The MEC-EABP interface was designed to: 1) amplify MEC volume displacement to achieve proper balloon inflation, 2) maintain a secure and comfortable anatomic fit, 3) optimize energy transfer efficiency, 4) meet muscle force and speed requirements, 5) balance work storage and delivery for rapid balloon inflation/deflation, 6) minimize tissue/device reactivity, and 7) maximize device durability. The eVAD was then prototyped and bench tested to assess its viability as a long-term cardiac assist device. Results showed that the manufactured MEC-EABP system meets all seven design criteria listed above, demonstrating the overall feasibility of this approach.

The cVAD represents an alternate approach to delivering muscle power via the MEC to boost cardiac output. Like the eVAD, this device supports the heart without touching the blood and so avoids the serious thromboembolic complications commonly associated with long-term

VAD use. Unlike the eVAD however, which unloads the left ventricle indirectly via aortic counterpulsation delivered during cardiac diastole, the cVAD is designed to compress the epicardial surface of both ventricles during the systolic portion of the cardiac cycle, thereby providing support to both sides of the heart.

Sleeve design was optimized via finite element analysis (FEA) simulations while biventricular deformations were simulated under various intra-ventricular and epicardial pressures to quantify the compression pressures required to achieve clinically significant improvements in cardiac performance. The sleeve material and manufacturing method were selected after a series of rigorous material testing and iterative prototyping processes. Results showed that a soft robotic sleeve 3D printed with ChronoSil meets all material and performance criteria for this application.

Ultimately, whether the chosen approach is counterpulsation EABP or copulsation DCCS, these muscle-powered systems serve to both reduce the risk of infection and enhance patient

vi quality-of-life by eliminating the need for external hardware components. Moreover, and of equal importance, using muscle power to actuate these non-blood-contacting pumps avoids thromboembolic events and obviates the need for long-term antithrombotic therapies. Therefore, these devices would, in principle, be a more attractive option for destination therapy as they would be simpler to maintain and hence less expensive in aggregate than traditional blood pumps, thereby resulting in wider availability and reduced costs for healthcare providers.

vii

Table of Contents

Acknowledgments……………………………………………………………………..….……..iii

Abstract………………………………………………………………………………….…..……v

Table of Contents………………………………………………………………………………viii

List of Tables………………………………………………………………………….….……..xiv

List of Figures………………………………………………………………………..………….xv

Chapter 1. Background and Significance ...... 1

1.1 The Need for Mechanical Circulatory Support ...... 1

1.1.1 Congestive Heart Failure ...... 1

1.1.2 Bridge Devices ...... 2

1.1.3 Destination Therapy ...... 3

1.2 The History of Cardiac Assist Devices ...... 3

1.2.1 The Beginning ...... 3

1.2.2 First Generation: Pulsatile Pumps ...... 4

1.2.2.1 LVAD ...... 4

1.2.2.2 RVAD ...... 5

1.2.2.3 BiVAD ...... 6

1.2.2.4 Total Artificial Heart ...... 6

1.2.3 Second Generation: Continuous Axial Flow Pumps ...... 7

1.2.4 Third Generation: Continuous Centrifugal Pumps ...... 8

viii

1.3 Current State of The Art ...... 10

1.3.1 CADs in Clinical Settings Today ...... 10

1.3.1.1 Short-term Circulatory Support ...... 11

1.3.1.2 CADs for Extended Use ...... 14

1.3.1.3 Pediatric Pumps ...... 15

1.3.2 Clinical Complications of Current VADs ...... 16

1.3.2.1 Driveline Infections ...... 17

1.3.2.2 Pump Thrombosis ...... 18

1.3.2.3 Gastrointestinal Bleeding ...... 19

1.4 Innovations for Effective Long-Term Cardiac Support ...... 20

1.4.1 Alternative Powering Methods for Untethered Cardiac Support ...... 21

1.4.1.1 Transcutaneous Energy Transfer System ...... 22

1.4.1.2 Muscle-powered VADs ...... 24

1.4.2 Non-Blood-Contacting Cardiac Assist Devices...... 25

1.4.2.1 Copulsation Direct Cardiac Compression Sleeve ...... 26

1.4.2.2 Counterpulsation Extra-Aortic Balloon Pump ...... 28

1.4.2.3 Passive Periventricular Restraint ...... 29

1.4.3 CADs in Summary ...... 30

1.4.4 Patient Management for Long-Term Treatment ...... 32

1.5 Project Relevance / Scientific Merit ...... 32

ix

Chapter 2. Implantable Volume Amplification Mechanism (iVAM) ...... 34

2.1 Foundation Technologies and Rationale ...... 34

2.1.1 Harnessing Skeletal Muscle as an Internal Power Source ...... 34

2.1.2 Extra-Aortic Counterpulsation ...... 36

2.2 iVAM Design Elements ...... 38

2.2.1 Volume Amplification ...... 38

2.2.2 Anatomic Fit ...... 39

2.2.3 Energy Transfer Efficiency ...... 40

2.2.4 Muscle Force and Speed Requirements ...... 43

2.2.5 Work Storage and Delivery ...... 46

2.2.6 Material Selection ...... 49

2.2.7 Device Durability ...... 50

Chapter 3. Extra-Aortic Ventricular Assist Device (eVAD) ...... 52

3.1 Prototyping and Assessment ...... 52

3.1.1 Test Bench Setup ...... 53

3.1.2 In Vitro Test Results for System Accuracy and Viability ...... 55

3.1.2.1 iVAM Volume Amplification ...... 55

3.1.2.2 Force Competency for Hypertensive Cases ...... 56

3.1.2.3 Speed Competency for Tachycardia ...... 58

3.1.2.4 Energy Distribution ...... 58

x

3.1.3 Does Our Prototype Meet Design Criteria? ...... 60

3.1.4 Limitations ...... 61

3.2 Surgical Approach and Implant Configuration ...... 62

3.3 Alternate Application ...... 63

3.4 Conclusions ...... 64

Chapter 4. Compressive Ventricular Assist Device (cVAD): Design ...... 65

4.1 Soft Robotic Direct Cardiac Compression Sleeve (DCCS)...... 65

4.2 Sleeve Geometry and Concept ...... 66

4.2.1 Geometry...... 66

4.2.2 Wall Thickness...... 69

4.2.3 Material Requirements ...... 69

4.2.4 Pressure Requirements ...... 70

4.2.5 Number of Tubes ...... 70

4.2.6 Manufacturing ...... 71

4.2.7 Implantation ...... 72

4.3 Sleeve Design Optimization via Finite Element Analysis ...... 73

4.3.1 Materials and Methods ...... 73

4.3.1.1 Biventricular Model Geometry ...... 73

4.3.1.2 Constitutive Modeling of Passive Heart Tissue ...... 74

4.3.1.3 Boundary and Loading Conditions ...... 78

xi

4.3.1.4 Mesh Cavity Extractions for SV and EF Analysis ...... 79

4.3.2 Geometric Deformation and Cardiac Output Results ...... 80

4.3.2.1 Uniform EP at Zero Afterload ...... 82

4.3.2.2 Separate EPs at Zero Afterload ...... 82

4.3.2.3 Uniform EP at End-Systolic Afterloads ...... 83

4.3.2.4 Separate EPs at End-Systolic Afterloads ...... 84

4.3.3 Stress, Strain, and Energy ...... 84

4.3.4 Independent LV and RV Epicardial Pressures...... 86

4.3.5 Limitations of the BiV Model Simulations ...... 88

4.3.6 Future Steps ...... 89

Chapter 5. Compressive Ventricular Assist Device (cVAD): Prototype and Assessment .... 91

5.1 Soft Material 3D Printing ...... 91

5.1.1 3D Printing Soft Robotic Actuators ...... 91

5.1.1.1 Additive Manufacturing of Flexible Materials ...... 92

5.1.1.2 3D Printable Soft Polymers ...... 93

5.2 Evaluation of Five 3D Printed Sleeve Prototypes ...... 95

5.2.1 Prototypability...... 95

5.2.1.1 Characterizations of Five Potential Material Options ...... 95

5.2.1.2 Print Settings...... 100

5.2.1.3 Two out of the Five Printed Prototypes Pass ...... 100

xii

5.2.2 Device Performance and Functionality ...... 103

5.2.2.1 In Vitro Test Setup and Methods ...... 103

5.2.2.2 Volume Amplification and Work Requirement ...... 106

5.2.3 Device Implantability...... 106

5.2.3.1 Lap Shear Test on Lamb Heart Epicardium ...... 107

5.2.3.2 High Frequency Cyclical Loading Test ...... 109

5.2.3.3 Biocompatibility ...... 113

5.3 Ex Vivo Implementation on Fetal Bovine Hearts ...... 113

5.4 Conclusion and Discussions ...... 116

Chapter 6. Summary and Conclusions ...... 119

References……………………………………………………………………………….….….124

xiii

List of Tables

Table 1. Commonly used cardiac assist devices and their key characteristics ...... 31

Table 2. The ∆P, TKE, and flow path profiles of different iVAM design iterations ...... 41

Table 3. Parameters used in the passive constitute law of myocardial fibers...... 76

Table 4. Summary table of LVEP, RVEP and total strain energy ...... 87

Table 5. Print settings and material properties of the five 3D printable soft polymers ...... 96

Table 6. Adhesion energies generated from lap shear testing ...... 109

Table 7. Summary table of the five 3D printable flexible polymers assessments ...... 116

xiv

List of Figures

Figure 1. First- and second-generation pumps ...... 8

Figure 2. Timeline of important milestones of CAD development history ...... 9

Figure 3. Examples of first, second, and third generation cardiac assist devices ...... 10

Figure 4. Examples of temporary support mechanisms ...... 13

Figure 5. Some of the most longstanding complications after LVAD implantations ...... 20

Figure 6. Schematics of the TET system ...... 23

Figure 7. Muscle-powered VADs could use the latissimus dorsi as its power source ...... 25

Figure 8. Examples of soft robotic direct cardiac compressive sleeves...... 29

Figure 9. Computer rendering of the MEC device...... 35

Figure 10. Intra- and extra-aortic balloon pumps...... 37

Figure 11. Free body diagram of the MEC-iVAM complex ...... 39

Figure 12. Full and cross-sectional views of the MEC-iVAM complex ...... 40

Figure 13. CFD of a one-eighth section of the MEC-iVAM complex design ...... 43

Figure 14. Schematics of the initial and final states of the MEC-iVAM complex...... 45

Figure 15. Force requirement and work distribution of the MEC-iVAM complex ...... 46

Figure 16. 3D CAD model of a complete muscle-powered counterpulsation system...... 50

Figure 17. Minimum flex life of the iVAM bellows exceeds 190M cycles ...... 51

Figure 18. MEC-iVAM complex manufactured and assembled...... 52

Figure 19. Bench setup schematics for the muscle-powered counterpulsation VAD...... 54

Figure 20. Bench setup for C-pulse EABP device actuation...... 55

Figure 21. Bench setup for EABP actuation against a pressurized mock silicone aorta...... 56

Figure 22. Peak MEC-iVAM actuation forces against different afterloads...... 57

xv

Figure 23. Complete eVAD system implanted in a human thoracic cavity...... 63

Figure 24. Computer rendering of a muscle-powered copulsation system in a thoracic cavity .. 66

Figure 25. Thin-walled tubes are arranged in a circle to form a soft robotic DCCS ...... 68

Figure 26. Stress and geometric deformation experienced by FEA models ...... 71

Figure 27. Computer rendering and 3D printed prototype of the current DCCS design...... 72

Figure 28. Reconstructed human biventricular models ...... 74

Figure 29. Stress-strain curve of the heart tissue data overlay with Yeoh 3rd model...... 77

Figure 30. Computer renderings of a biventricular model undeformed and deformed ...... 78

Figure 31. Stroke volumes and ejection fractions of the BiV model ...... 80

Figure 32. BiV model deformations after cardiac compressions ...... 81

Figure 33. RVEP-to-LVEP ratio for zero and end-systolic afterloads ...... 83

Figure 34. Stress, strain, and energy experienced by the BiV model ...... 85

Figure 35. Five types of nonrigid Type IV specimens were 3D printed and tested ...... 99

Figure 36. Cylindrical sleeve actuator was designed and 3D printed using five materials...... 102

Figure 37. Test bench setup and reuslts of the FormLabs and ChronoSil prototypes...... 105

Figure 38. Material adherence testing on lamb heart epicardium strips...... 108

Figure 39. High frequency cyclical loading on a ChronoSil sample for durability testing ...... 112

Figure 40. Ex vivo cardiac compression demonstration on a fetal bovine heart ...... 115

Figure 41. Schematic of the bench setup for cVAD system assessment...... 118

xvi

Chapter 1

Background and Significance

1.1 The Need for Mechanical Circulatory Support

1.1.1 Congestive Heart Failure

Congestive heart failure (CHF) is a progressive condition in which cardiac function deteriorates over time. It is most common among people 65 years or older, but practically anyone can be at risk as the causes of heart failure include everything from coronary artery disease, high blood pressure, and congenital heart defects to myocarditis, abnormal heart rhythms, valve disease, diabetes, and obesity. The most common symptoms of the disease include shortness of breath and fatigue, and it is often diagnosed via blood tests, electrocardiograms, echocardiograms, stress tests, coronary angiograms, and chest x-rays.1 CHF remains one of the most costly diseases in the industrialized world, both in terms of healthcare dollars and the loss of human life. It is estimated that 26 million people currently suffer from CHF worldwide, including 5.8 million people in the

United States where the economic impact exceeds $40 billion per year in medical costs and lost productivity. Worse still, roughly half of all people who develop CHF die within five years of diagnosis due to the limitations of current long-term treatment strategies.2,3 Cardiac transplantation

1 is generally considered to be the best recourse for end-stage CHF patients, but this treatment option is not available to most patients as the number of donated hearts is restricted by roughly 2,200 hearts per year in the U.S.4 Pharmacologic therapies can improve heart function in the short term and relieve the symptoms associated with CHF, but are unable to restore and maintain normal heart function over the long term.5,6 Therefore, decades of development work have focused on cardiac assist devices (CADs) as an alternate solution for end-stage CHF patients.

1.1.2 Bridge Devices

CADs are often categorized according to duration of support. If a device lasts from hours to weeks as a means to stabilize patients until longer-term mechanical support can be implemented, it is considered to be a ‘bridge-to-device (BTD)’.7 BTDs were commonly used for myocardial recovery and mitral valve replacement from the 1970s through early 1980s. Nowadays, only about

25% of cases use this temporary treatment option, typically for one to four weeks post-operation.8

CADs may also be used to provide circulatory support to patients on the waiting list for heart transplantation, in which case they are considered to be ‘bridge-to-transplant (BTT)’ devices.

While BTT devices do not last longer than 2 years on average, the longevity of CADs is much better today compared to that of the mid-80s where the typical working life of these devices was only about a month.9 Yet, this improvement does not increase the total number of patients who can receive a transplanted heart, but rather increases the chances of receiving a transplanted heart for patients receiving CAD treatment while proportionally reducing the odds for those who do not.9,10

In rare instances that are difficult to predict, some patients recover cardiac function while under CAD support and no longer need heart transplantation. These cases are categorized—more often in retrospect than by design—as ‘bridge-to-recovery (BTR)’.11

2

1.1.3 Destination Therapy

Currently, the most ambitious unmet goal in the CAD field is to develop a cardiac support system for long-term or permanent use. A safe, reliable, durable, implantable support mechanism leading to the preservation, or even restoration, of cardiac competence and coronary flow that completely frees patients from the need for heart transplantation would be considered an effective device for ‘destination therapy (DT)’.12 In order to achieve this goal, researches have focused on overcoming several major failure modes associated with extended circulatory support. In this paper, we review the historical efforts, contemporary technologies, and up-to-date cutting-edge innovations that have been made to develop durable and reliable devices that both support cardiac function for long-term survival and also provide for better patient quality-of-life.

1.2 The History of Cardiac Assist Devices

1.2.1 The Beginning

The concept of artificial blood pumps can be traced as far back as 1813 when Le Gallios first performed the task by squeezing rubberized pumping chambers between pairs of wooden planks.13 But it was not until the 1960s when cardiac assist devices finally began to replace cardiopulmonary bypass circuits as a means to support the failing heart.14 The earliest mechanical assist devices were pneumatically driven. The first implantable artificial ventricles in clinical use was reported by Liotta in 1963 and it consisted of a pneumatically-compressed valved tubular conduit that connected the left atrium to the descending aorta.14 A double-lined restraint cup that wrapped around the ventricles and alternately inflated and deflated to displace blood from both ventricles (reported by G. Anstadt and P. Schiff in 1966) was also a pneumatic device.15 An air- powered balloon pump that provides effective left ventricular unloading and systemic circulatory

3 support by displacing blood from the descending aorta during the diastolic phase of the cardiac cycle was first used clinically in 1968.16,17 Around the same time, the idea of complete replacement of the entire heart using a pneumatic total artificial heart (TAH) emerged and the implantation procedure was first performed clinically in 1969.18,19 However, because these early attempts risked a high rate of fatality from sudden device failures, focus shifted toward the use of simpler single- chambered mechanical blood pumps for univentricular support, known as ventricular assist devices (VADs).3,18

1.2.2 First Generation: Pulsatile Pumps

When VADs were first developed, they were designed to replicate the native cardiac cycle and generate pulsatile flow using a diaphragm and unidirectional artificial heart valves (Figure

1A).3 The first generation VADs were either pneumatically or electrically driven and included larger pulsatile VADs like HeartMate XVE (Thoratec, Pleasanton, CA, USA) and Berlin Heart

EXCOR (Berlin Heart, Berlin, Germany) that were used to support patients awaiting cardiac transplantation.18–21 These earlier pulsatile pumps were characterized by their large size, heavy weight, and an external driving unit that seriously limited patients mobility. These first generation pulsatile VADs could be used either as a left ventricular assist device (LVAD), a right ventricular assist device (RVAD), or as a biventricular assist device (BiVAD).

1.2.2.1 LVAD

Because the left ventricle (LV) and right ventricle (RV) can be supported either separately or in unison, ventricular assistance is commonly separated into LVAD, RVAD, and BiVAD categories.22 With isolated LVAD therapy, the systemic circulation is typically supported by

4 drawing blood from the left ventricular apex and pumping it into the ascending aorta. This not only restores perfusion to all organs and tissues outside the pulmonary circulation (including the heart itself), but also unloads the LV, which may prevent or even reverse pathologic LV remodeling caused by chronic pressure overload. Subsequent effects on RV function are complex however, as right-side improvements resulting from lower pulmonary pressures are offset by several factors that could lead to RV failure, including: increased preload, leftward shift and reduced contractility of the interventricular septum, increased work demand to match LVAD output, and tricuspid valvular distortions. The first successful LVAD implantation was completed by De Bakey in 1966, and the majority of cardiac support research has been dominated by LVAD developments for clinical practice ever since. Some first generation pulsatile LVADs include Novacor LVAS

(Baxter Healthcare, Oakland, CA, USA), HeartMate I (Thoratec), and Thoratec PVAD

(Thoratec).23–26

1.2.2.2 RVAD

The clinical settings in which RVAD therapy are most commonly employed include acute myocardial infarction, pulmonary embolism, pulmonary hypertension, myocarditis, post- cardiotomy shock, cardiac transplantation, and LVAD implantation. As the frequency of LVAD use continues to rise, this last scenario is becoming increasingly common as nearly half of all CHF patients show right heart failure after LVAD implantation and 4% require RV support within the first two weeks post-operation.27,28 Because RV complications after LVAD surgery are both relatively frequent and highly significant in terms of morbidity and mortality, the means to provide right ventricular mechanical support is now considered an essential capability in medical centers where LVAD therapy is performed.18,28 Today, some RVADs like SynCardia (SynCardia Systems,

5

Tucson, AZ, USA) serve as BTT while some others like Impella RP (AbioMed, Danvers, MA,

USA), TandemHeart (CardiacAssist, Pittsburgh, PA, USA), and CentriMag RVAD (Thoratec) serve as peri-operative bridges to mechanical support.18,29

1.2.2.3 BiVAD

While the majority of patients retain sufficient RV function throughout the course of

LVAD therapy to avoid the need for ancillary support, nearly 48% of LVAD recipients experience sufficient levels of postoperative RV dysfunction to warrant the use of a biventricular assist device.27 BiVAD is especially helpful for patients with total heart failure because it supports both sides of the failing heart by balancing left and right pump flows and, in rare cases, inducing myocardial recovery. The first generation pulsatile BiVADs have saved many lives, but are limited by their bulkiness, the necessity of a large external pneumatic driver that inhibits patient mobility, infection at the driveline site, and thrombus formation. Some first generation BiVADs include

AbioMed BVS5000 (AbioMed), Berlin Heart EXCOR (Berlin Heart), and Medos HIA-VAD

(Stolberg, Germany).27,30

1.2.2.4 Total Artificial Heart

Total artificial hearts are designed to entirely replace native heart function over extended periods to treat end-stage CHF. The first human TAH implantation was performed in 1969 by

Denton Cooley using the Liotta artificial heart as a bridge to cardiac transplantation. The patient was supported on this pneumatic device for three days during which time hemolysis and deteriorating renal function prompted surgeons to replace the pump with a donor heart that failed

36 hours later.18,19 It was not until 1982 when the Jarvik-7 TAH (Jarvik Heart, New York, NY,

6

USA) was able to support a patient for 112 days that these devices were generally considered a viable means to support patients for BTT.19 CardioWest (SynCardia), which the Jarvik 7 later became, and Abiocor (AbioMed) are examples of TAHs that have been used clinically.31

1.2.3 Second Generation: Continuous Axial Flow Pumps

Because first generation pulsatile pumps were limited by their large size, high noise emission, and durability issues leading to frequent malfunction and morbidity, research to develop smaller and more reliable devices were initiated and continued through the 1990s.17 As a result of this work, Thoratec introduced a new VAD in 2001 called HeartMate II that was just one-seventh the size and one-quarter the weight of the original HeartMate XVE.20,21 This radical design change was achieved by integrating a valveless axial pump with a variable magnetic field designed to rapidly spin a single impeller that produces continuous outflow directed in parallel to the axis of rotation (Figure 1B).3 HeartMate II received FDA approval for BTT in 2008 and for destination therapy in 2010.32 To date, over 26,600 patients have received HeartMate II LVAD demonstrating

85% survival at one year.33 Other axial flow pumps developed during this same time period included Hemopump (Medtronic), DeBakey VAD (Micromed), HeartAssist-5 (Reliant Heart,

Houston, TX, USA), Jarvik 2000 (Jarvik Heart), Impella (Abiomed), and Incor (Berlin Heart).

These second generation LVADs were able to provide patients with a better quality of life, mobility, and restoration of heart function compared to the first generation positive displacement

VADs, but still relied on extracorporeal power sources and required patients to undergo constant anticoagulation therapy for the duration of the implant due to the risk of thromboembolic events.18

7

Figure 1. The first generation pulsatile-flow pumps (A) replicated the native cardiac cycle using a diaphragm and unidirectional artificial heart valves, while the second generation continuous- flow pumps (B) integrated a valveless axial pump designed to rapidly spin a single impeller.

1.2.4 Third Generation: Continuous Centrifugal Pumps

The third generation LVADs are continuous flow centrifugal pumps designed with magnetic and/or hydrodynamic levitation of the impeller with non-contact bearings and its outflow directed perpendicular to the axis of rotation.3,34 These radial rotary pumps feature further reduced device size, noise emission, infection rate, and prothrombotic sites for better patient outcomes and lifestyles.18 Now that nearly 99% of LVADs placed are continuous flow LVADs (CF-LVADs) today, third generation centrifugal pumps such as HeartWare HVAD (HeartWare), HeartMate III

(Thoratec), CentriMag (Thoratec), Incor (Berlin Heart), Levacor (World Heart, Salk Lake City,

UT, USA), and DuraHeart (Terumo Heart, Ann Arbor, MI, USA) play big roles.3,26 HeartWare

HVAD and HeartMate III received FDA approval for long-term mechanical circulatory support in

2017 and 2018, respectively, and CentriMag was approved to support one or both sides of the heart for up to 30 days in patients.35–37 Some other milestones in VAD development history are summarized in the timeline shown in Figure 2, while some of the most popularly used first-, second-, and third-generation VADs are illustrated in Figure 3. Despite significant improvements in device function and durability, however, complications like right heart failure, infection, thrombosis, hemolysis, and neurologic events still persist.16

8

Figure 2. Timeline of important milestones of cardiac assist device (CAD) development history.38,39

9

Figure 3. Examples of first, second, and third generation cardiac assist devices.40–46

1.3 Current State of The Art

1.3.1 CADs in Clinical Settings Today

After five-plus decades of dedicated research aimed at developing blood pump technologies to support the failing heart, a cadre of devices capable of delivering different levels of support at different levels of invasiveness are now available to treat different varieties and severities of cardiac malfunction. These range from acute catheter-based interventions used for partial univentricular support to long-term implantable pumps designed to restore normal perfusion

10 levels in both systemic and pulmonary circulations.3,18,20 Although guidance on patient selection for mechanical support is limited, the criteria usually include a combination of factors such as patient age, body size, cardiac malfunction type, disease stage, and candidacy for organ transplantation. For example, patients who require immediate VAD replacement due to the severity of their symptoms and/or are expected to have longer than normal wait times on the transplant list due to their body size and blood type are generally considered to be candidates for

BTT devices. Alternatively, patients who require circulatory support but for some reason cannot be—or do not wish to be—listed for cardiac transplantation surgery are treated as destination therapy candidates.47,48

1.3.1.1 Short-term Circulatory Support

Extracorporeal membrane oxygenation (ECMO) (Figure 4A) is a form of cardiopulmonary bypass that is used as a bridge to recovery, transplantation, or mechanical circulatory support.49 It provides blood oxygenation and circulation with a mechanical pump stationed outside the body.

ECMO is generally used in an emergent setting and continued until symptoms are improved, but the typical time course is hours to days because long-term ECMO support increases the likelihood of thrombotic complications.49,50 Even though ECMO has been in clinical use as a class II/III device for over 30 years, the decision to use it remains a risk vs. benefit situation because complication rates are high as occurrences of bleeding and infection reach up to 40% and 31%, respectively. Patients with neurologic injuries, hemorrhage, immunosuppression and/or advanced age are generally thought be poor candidates for ECMO treatment.51,52

AbioMed’s Impella catheter (Figure 4B) is an intravascular microaxial blood pump that provides partial circulatory support from a few hours to one month maximum.53,54 Left ventricular

11

Impella catheters come in three different models: Impella 2.5, Impella CP, and Impella 5.0, which produce flow rates up to 2.5 /min, 3.5 L/min, and 5.0 L/min, respectively. All three are designed to circulate blood by placing their inlet in the LV and outlet in the ascending aorta. Similarly, there is Impella RP designed for partial right sided circulatory support, which provides up to 4.0 L/min of blood flow to the pulmonary circulation. Just last year in 2018, Impella Ventricular Support

System received approval for expanded FDA indications for cardiomyopathy and percutaneous coronary intervention procedures after demonstrating its safety and effectiveness on over 50,000 patients treated from 2008 to 2017.55–57 One major caveat with these devices is that their proper function is highly dependent on the correct position of the catheters, which makes post-implant management of these catheter-based pumps critically important. All models come with an

Automated Impella Controller (AIC) that monitors and controls the overall system.53,57,58

Pneumatic intra-aortic balloon pumps (IABP) (Figure 4C), which are internal counterpulsation devices placed inside the descending aorta, has been one of the most common mechanical support systems for the failing heart ever since it was classified as a class III device in

1979.59,60 The balloon is inflated during ventricular diastole to increase diastolic pressure, coronary blood flow, and systemic perfusion, and rapidly deflated during systole to induce reduced cardiac afterload and enhanced cardiac output. The IABP is actually one of the earliest CADs developed, with the first preliminary studies done as early as 1961 by Kanitrowitx and Moulopoulos and the first successful clinical application reported in 1967.59,61 Most IABPs in clinical use today are predominantly Arrow IABP series, now acquired by Teleflex Medical. Because proper actuation timing is crucial for counterpulsation therapy, Teleflex Arrow IAB Catheters come with their own

AutoCAT2 control unit that has both AutoPilot Mode, which automatically selects appropriate settings using arterial pressure waveforms as the guideline, and Operator Mode in which all

12 settings are user-controlled. The catheter balloons also come in different sizes for different sized patients.61

Thoractec’s CentriMag acute circulatory support system (Figure 4D), a temporary external

VAD that can support right, left, or both ventricles, was the first and only magnetically levitated blood pump cleared by FDA in 2008.62 It is a continuous flow centrifugal pump without bearings or seals that operates at speeds up to 5500 rpm, delivering up to 9.9 L/min blood flow for a maximum recommended support duration of 6 hours.63 This short-term solution for acute heart failure features a magnetically levitated pump impeller that operates within a contact-free environment to help minimize blood-related complications. The CentriMag system comes with a pump, a motor, a console with dual display monitor, a back-up console battery with a 5-hour recharge time, and a power conditioning unit that is air transport operable with AC power and able to accommodate up to four CentriMag consoles simultaneously.63

Figure 4. ECMO (A), AbioMed Impella (B), Teleflex Arrow IABP (C), and Thoratec CentriMag (D) are temporary support mechanisms commonly used in clinical settings today.64–67

13

1.3.1.2 CADs for Extended Use

Although the Heartmate II axial flow pump remains the world’s most widely used and extensively studied VAD to date with over 26,000 patients implanted for periods up to 10 years and beyond, third-generation centrifugal pumps like the HeartMate III (Thoratec) and HeartWare

HVAD (HeartWare) are currently poised to become the device of choice as either BTT or DT for end-stage CHF patients. HeartMate III (Figure 4) was built upon the HeartMate II platform but with key improvements that include a modular driveline, mobile power unit interface, no surgical pocket, less power consumption, and most importantly, a unique magnetically levitated core system called Full MagLev technology.68 This proprietary maglev system reduces overall blood trauma and maximizes hemocompatibility by maintaining large and consistent gaps within the pump housing and features an optional pulse mode as a means to minimize stasis and provide pulsatile flow to perfused organs. This design allows for significantly less shear stress (hemolysis) and blood-contacting surface area (thrombosis) since the size of the flow path that allows red blood cells to pass without rotor-housing contact is more than 20 times larger than that of its predecessor.68 In addition to the pump itself, the HeartMate III system comes equipped with an external controller that powers and checks the pump and driveline, a percutaneous driveline, and an external battery pack. Refined implantation techniques together with improvements in mechanical reliability, pumping efficiency, and battery life have increased 2-year survival rates from 76.2% to 82.8% while also contributing to surgical ease and patient quality-of-life.69,70

HeartWare HVAD is a small CF-LVAD with a displacement volume of 50 ml and an output capacity of 10 L/min.71 It is characterized by a unique wide-blade impeller and a hybrid magnetic- hydrodynamic suspension technology that ensure no mechanical contact within the pump and a dual-motor system that is designed for increased efficiency and reliability.34 It comes with a rotary

14 pump that operates at speeds ranging from 1800 to 4000 rpm, a percutaneous driveline, an external microprocessor-based controller, a monitor that displays and logs downloadable waveform data, lithium-ion batteries that allows patient mobility for about 4 to 6 hours, AC/DC power adapters, and a battery charger. Its small device size and cannula allow minimal invasiveness and therefore faster postoperative recovery and better clinical outcomes.34,71

SynCardia CardioWest TAH (SynCardia) is the world’s first and only commercially approved total artificial heart that is currently in use today.72 The mechanics of the device are fairly simple. It delivers pulsatile flow up to 9L/min by filling two artificial ventricles that are sutured to the patient’s aorta and pulmonary artery and ejects blood through unidirectional valves via a pneumatically driven diaphragm.31 According to INTERMACS reports, SynCardia TAH recipients experienced significantly fewer neurologic and thromboembolic events compared to

BiVAD recipients.31 It has notably increased patients support time and currently has an overall one-year survival rate of 67.6%.73

1.3.1.3 Pediatric Pumps

Conventional continuous flow VADs were designed specifically to treat adult patients, who comprise the vast majority of the end-stage CHF population and so tend to be too large for use in pediatric patients weighing less than 25 kg (55 lbs.).74 Berlin Heart EXCOR Pediatric is a pulsatile paracorporeal VAD designed for left and/or right ventricular support of young patients from newborns to adolescents.75 It is composed of a cannula that comes in different tip types and sizes, a blood pump that also varies in sizes from 10 to 60 cc, and a driving unit that provides alternating pneumatic pressures. The system can be powered by either the stationary IKUS driving unit or a portable battery unit that lasts for roughly 6 hours. To monitor patients, the IKUS unit is integrated

15 with laptop software that is programmed to log and store data as well as alarm both visually and audibly when waveform readings are abnormal.75 Besides Berlin Heart EXCOR, other pediatric pumps or miniature adult pumps include the Jarvik 2000 (Jarvik Heart) that come in different sizes for children and infants, PediaFlow (PediaFlow Consortium, Pittsburgh, PA, USA) that supports infants and young children weighing 2–25 kg, the miniature MVAD HeartWare (HeartWare), and

CircuLite (CircuLite Inc., Saddle Brook, NJ, USA).76

1.3.2 Clinical Complications of Current VADs

In spite of the increasing number of VAD options currently available to patients due to revolutionary advances in cardiac support technologies, numerous challenges still persist.

Ventricular arrhythmia, right heart failure, infection, pump thrombosis, and bleeding are still areas of concern, as are issues of long-term patient management and a lack of clear guidelines regarding patient eligibility criteria for VAD therapy.77 Difficulties in gauging the likelihood of therapeutic benefit for any given individual heart failure (HF) patient is thought to be the biggest reason behind the recent plateauing of VAD use. Optimization of the treatment process and refinements in patient selection criteria are therefore needed to promote further improvements in survival rate and patient quality of life, especially in the setting of long-term circulatory support.

Indeed, given that heart failure has now risen to pandemic proportions across the globe while the availability of donor hearts remains woefully inadequate to meet the rising demand, continued expansion of mechanical circulatory support for use as long-term BTT or DT is considered a clinical necessity. But despite decades of development most VAD therapies are limited to short-term BTT applications due to three longstanding complications. One is bacterial infection from percutaneous drivelines, which is the most frequent LVAD-associated problem.5

16

Another is thromboembolic events associated with blood-contacting surfaces, which includes both pump thrombus formation and blood clotting in the circulatory system.3,6 And the third is bleeding, mainly at the surgical site during the early postoperative period and gastrointestinal bleeding that usually begins three months after continuous flow LVAD implantation.78

1.3.2.1 Driveline Infections

Device malfunction, bleeding, thrombosis, and inadequate aftercare all contribute to VAD failure in the clinical setting, but percutaneous driveline infection (DLI) (Figure 5A) is one of the most common cause of mortality with these devices, accounting for 47% of all unplanned readmissions for LVAD patients.18,30 This risk factor has proven difficult to avoid in these pumps as drivelines that provide power, control, and communication are percutaneously sutured to remain secure, and this driveline exit site creates a conduit for bacterial entry that often leads to DLI. The prevalence and seriousness of DLI, which often leads to erythema, hyperthermia, purulent drainage, and significantly lower survival rate, increased as LVAD therapy expanded from short- term to long-term use.3,16 Although approximately 70% of infected patients require rehospitalization in the first year, there currently is no comprehensive guideline for DLI treatment besides general precautions like minimal exit-site movement, long-term suppressive antibiotics, and antimicrobial therapy.20,21,79,80 Ongoing efforts to decrease DLI incidents include optimization of driveline implantation techniques and minimization of pump profile and operational invasiveness, which has resulted in smaller and more efficient devices such as the entirely intra-pericardial HVAD

(HeartWare) and completely intra-thoracic HeartMate III (Thoratec).16 However, in all cases a tunneled percutaneous driveline is still required for power delivery from sources outside the body.17

17

1.3.2.2 Pump Thrombosis

Another significant cause of LVAD complications is thromboembolism (Figure 5B) associated with blood-contacting surfaces.3,6 Pump thrombosis, where blood clots form at the blood-device interface, is a multifactorial process caused by misuse of anticoagulants, abnormal angulation of cannulas, and surface mediation of blood-contacting devices.3 Thrombosis can occur in any component of the LVAD in contact with the bloodstream and may result in turbulent flow, elevation in device power consumption and, in extreme cases, inability to unload the LV.21 The annual incidence of pump thrombosis in LVAD patients exceeds 10%, of which nearly one third lead to serious complications including aortic insufficiency, hemolysis, neurologic events, and cardiogenic shock.20,21 From the time of confirmed pump thrombosis, there is a two-fold increase in mortality at 30, 90, and 180 days, where mortality reaches 48.2% if no LVAD exchange or cardiac transplantation is performed within that given time.20,21 This potential complication, common to all blood-contacting devices, requires VAD recipients to undergo costly—and potentially dangerous – anticoagulation therapy for the duration of the implant period. In order to minimize the rate of chronic pump thrombosis, innumerable changes in VAD designs have been made over the years. Modern LVAD surface area has been scaled down, impeller profiles have been adjusted, implantation invasiveness has been minimized, and less reactive surface materials have been chosen. Nonetheless, the risk persists and long-term antithrombotic therapies including anticoagulant drugs, antiplatelet agents, and routine surveillance are still required by patients receiving VAD therapy.20,21

18

1.3.2.3 Gastrointestinal Bleeding

The reported incidence of gastrointestinal bleeding (GIB) (Figure 5C) after continuous flow LVAD implantation is alarmingly high, as much as 61% by one account.81,82 There are several factors leading to GIB syndrome, but with third-generation continuous flow pumps the low pulsatility flow profile combined with increased oxidative and shear stresses seem to cause hematological abnormalities such as platelet dysfunction and von Willebrand factor (vWF) degradation.21,81 And chronic anticoagulative treatments like warfarin and antiplatelet agents like aspirin administered to prevent clot formation at blood contacting surfaces only worsen the risk of bleeding.21 GIB can be initially diagnosed and evaluated with endoscopy, but the most appropriate method of treatment after diagnosis is not always clear due to difficulties identifying the causes that underlie this complicated syndrome. Currently, a multidisciplinary approach that considers the location and severity of bleeding and thrombosis simultaneously is being used to manage GIB, but better insights into the etiology and treatment of GIB are still being sought to improve outcomes.81,82

19

Figure 5. Some of the most longstanding complications after left ventricular assist device (LVAD) implantations are bacterial infections at driveline entry site (A), pump thrombosis (B), and gastrointestinal bleeding (C).83–85

1.4 Innovations for Effective Long-Term Cardiac Support

The rate of DLI, thromboembolic incidents, and bleeding problems must be curtailed if long-term cardiac support is to become a viable treatment option for end-stage CHF patients.

Toward that end, there have been numerous attempts to eliminate these predominant failure modes and develop an untethered, non-blood-contacting VAD as a destination therapy.

20

1.4.1 Alternative Powering Methods for Untethered Cardiac Support

To provide long-term CAD patients better quality-of-life, various powering methods have been proposed to minimize or eliminate extracorporeal power requirements that limit patient autonomy and contribute to patient stress over potential power delivery failures (e.g., driveline fracture and battery exhaustion) and DLI risk. One of the most interesting attempts was a nuclear- powered device from the 1980s that used Plutonium-238 as a power source. The potential was in the nuclear radioisotope Plutonium that offered the highest possible energy density and long half- life without requiring any energy storage. However, the critical problems of heat dissipation and safety concerns regarding nuclear element leakage eventually led to termination of the project.86,87

Another attempt to develop a permanently implanted circulatory support system was based on a small, lightweight spring decoupled C-core solenoid that was first introduced in the early

1980s. This solenoid drive system was used to actuate a pair of preloaded beam springs that directly coupled to a dual pusher-plate blood pump, producing high starting forces and constant pump pressures through repeated ejection strokes. With this technology operating in combination with external plug-ins that included portable and rechargeable lithium-ion, nickel metal hybrid, or lead acid gel batteries, a level of patient mobility similar to that afforded by battery-powered devices available today was achieved.88,89 However, this drive system was limited not only by requiring patients to carry around external hardware like charging stations, battery packs, and emergency back-up systems in a backpack, but also by the anxiety produced by having to charge batteries every few hours.88,89 When findings from device malfunction cases were reviewed, researchers found that there were significant numbers of hospital visits due to device alarm of unknown origin and/or actual malfunctions resulting in controller exchange or battery change.

Although not all cases represented serious clinical complications, device alarms and malfunction

21 notices caused severe levels of anxiety and considerably reduced patient quality of life.90 Much worse, in some cases, patients actually died from battery exhaustion because of unexpected events that drained the batteries before they could be recharged.91 Therefore, alternative power sources for untethered pump operation have been sought to create totally implantable devices that are safe, reliable, and relatively maintenance-free.

1.4.1.1 Transcutaneous Energy Transfer System

Transcutaneous energy transmission (TET) technology (Figure 6A) that transfers power across intact skin makes devices completely implantable and therefore free of the risk of DLI.17 At a time when over 20 million Americans are estimated to have some type of implanted medical device, the TET system sounds extremely appealing.92 The idea of an inductive coupling of two coils that transfers electromagnetic energy at radio frequencies across a closed chest wall was first described by Schuder and colleagues in 1961.93 Because VADs tend to demand a higher range of power (up to about 25 W) compared to other implants like pacemakers or implantable cardiac defibrillators, transmission efficiency and the total amount of transferrable power are key performance criteria. Different methods of transmitting energy across skin such as ultrasonic energy transfer and acoustic energy transfer have been previously developed, but because inductive coupling TET outperforms the others by more than double in terms of efficiency, the latter has been used in devices like AbioCor TAH (AbioMed), which was FDA approved as a permanent TAH for humanitarian uses in 2006, and the LionHeart LVAD (Arrow International,

Reading, PA, USA), which received FDA approval for Phase I human clinical trials in 2001.93–98

The inductive electromagnetic TET system (Figure 6B) used in these devices has proven to be a promising wireless powering method that sufficiently meets the power transmission

22 requirement of up to 25 W.92,99 When studied with 14 AbioCor TAH patients, 30-day survival rate was 71% with no device-related infections reported, which clearly demonstrated the value of the

TET system regarding the elimination of DLI risks.93 However, this tether-free system is significantly limited by its power transmission range since the transmit and receive coils must remain very close together (within a few millimeters). This proximity restriction requires the receive coil to be implanted just under the skin and the external transmit coil to be secured in a single position on the skin surface with an adhesive dressing.93 The two coils can be distant for a very brief period of time (about 30 minutes), allowing activities like a brief shower.96 Another limitation is its lower energy efficiency compared to conventional extracorporeal drivelines as the

TET system consumes approximately 20% of the generated power during operation.93 Other drawbacks like fatal component failure, bleeding, and pain due to the large cumulative volume of all implanted parts also play a big role in preventing TET technology from being the main VAD powering method today.3

Figure 6. Schematics of the TET system A) in patient use and B) with an electromagnetic coupling between the internal and external coils located inside and outside of patient skin, respectively.100,101

23

1.4.1.2 Muscle-powered VADs

The use of electrically stimulated skeletal muscle as an endogenous power source to drive circulatory support systems is another alternative that is currently under study. An internal muscle energy converter that operates by converting the contractile energy of a muscle into a hydraulic power source would greatly simplify cardiac implants by eliminating electromechanical components and avoiding the need to transmit energy across the skin.102,103 A device powered by contractile energy and controlled via a pacemaker-like device implanted beneath the skin could, in principle, provide a safe, tether-free means to support the failing heart over extended periods of time.

The concept of muscle-powered cardiac support is not new. The use of untrained skeletal muscle to aid the failing heart dates way back. In 1935, Beck and Tichy employed static muscle grafts to revascularize the myocardium.103 And in 1958, Kantrowitz isolated diaphragm muscles in dogs to form pouches for use as ‘myocardial substitutes’.104 But in 1969 the concept of muscle- powered cardiac assist was given new life when Salmons and Jarvis demonstrated that myofiber properties can be changed from glycolytic fast type to oxidative slow-phenotype via muscle impulse activity training.104 This key discovery opened a whole new realm of possibilities involving conditioning skeletal muscle to provide fatigue-resistant long-term circulatory support.

The recent development of a functional muscle energy converter (MEC), which operates by converting endogenous muscle energy into hydraulic power, may ultimately provide CAD developers with the means to harness the body’s own energy to assist the failing heart over the long term.102,104 Among the several large skeletal muscles that might conceivably be used for this purpose, the MEC targets the latissimus dorsi muscle (LDM) (Figure 7A) due to its large size, surgical accessibility, proximity to the thoracic cavity, and steady-state work capacity sufficient

24 for long-term cardiac support.105 Trained LDM controlled by a programmable pacemaker-like cardiomyostimulator that coordinates muscle activity with the cardiac cycle has been shown to produce mechanical power at levels sufficient for pulsatile VAD actuation.103,104 As the current

MEC (Figure 7B) has been optimized to operate at contractile force and velocity levels that correspond to peak power generation in fully-conditioned human adult LDM, its potential as a means to power a completely self-contained VAD (Figure 7C) for long-term use is promising.102

Figure 7. Muscle-powered VADs could use the latissimus dorsi (A) as its power source and convert this endogenous muscular power into hydraulic energy via a completely implantable muscle energy converter (B) that can potentially power pulsatile VADs for long-term use (C).103,106,107

1.4.2 Non-Blood-Contacting Cardiac Assist Devices

Despite innumerable CAD designs and material modifications made over the decades in an attempt to eliminate chronic pump thrombosis, the situation still persists while the precise dosage and frequency of long-term antithrombotic therapies remain ambiguous.20,21 Consequently,

25 several groups are currently working to avoid this problem by designing non-blood contacting devices.3 These devices are intrinsically pulsatile and can be programmed to deliver energy to the bloodstream during cardiac systole (copulsation) or diastole (counterpulsation). Copulsation enhances cardiac output by increasing pulse and arterial pressure during systole, while counterpulsation boosts heart function by reducing aortic pressures as the heart fills thereby providing lower cardiac afterload for the failing heart.3 These techniques have been shown to significantly increase aortic peak pressure, cardiac output, and regional and coronary blood flow.108 But, above all, the most critical advantage these technologies offers is that they can be applied without touching the blood stream.

1.4.2.1 Copulsation Direct Cardiac Compression Sleeve

A normal heart with a ventricular ejection volume of about 71.5 mL per beat (CO = 5 L/min and HR = 70 bpm) has a ventricular ejection fraction (EF) of 60%. While a healthy heart’s EF ranges from 55% to 70%, anything less than that is considered mild (<54%) to severe (<35%) heart failure. One way to boost the EF of a defective heart is by applying pulsatile pressure to the epicardial surface in synchrony with the natural ventricular contraction.

Copulsative biventricular compression devices have been around for decades. The Anstadt

Assistor Cup became the first successful direct cardiac compression sleeve (DCCS) in 1991 and Dr.

DeBakey’s pneumatic LV compression cup was first implanted in 1996.20,21,109 As these preliminary ventricular DCCSs showed successful increases in arterial pressure and cardiac output, more pneumatic and electric sleeves were developed including the “cuff-like” Heart Booster (AbioMed) that covers and compresses the heart with parallel compression tubes110, Mannequin (Chase Medical,

Richardson, TX, USA) that restores round-shaped ventricles to its original oval-shape111, and Heart

26

Blanket (Leeds University, UK) that gives underperforming hearts an extra boost by contracting ventricles with piezoelectric bands in synchrony with pacemaker stimulations.112

Recently, researchers have turned to emerging soft robotic technologies to improve the long-term functionality of DCCSs. In 2017 for example, a silicone molded sleeve (Figure 8A) that employs McKibben pneumatic artificial muscles (PAMs) placed helically and circumferentially to both compress and twist the heart without contacting blood gathered a lot of attention.113 This soft robotic sleeve made of two biomimetic layers of contractile elements that shorten when pressurized during ventricular systole was able to restore cardiac output to 88% of normal when tested on porcine hearts.113 CorInnova’s minimally invasive soft robotic DCCS (Figure 8B) is a collapsible self-deploying device that wraps around the ventricles with custom fit thin-filmed pneumatic chambers. They were able to increase cardiac output by up to 50% in large animal acute heart failure studies.114 Unlike these pneumatic devices that are tethered to an external air supply, a muscle-powered DCCS (Figure 8C) that uses the geometric advantage produced by an array of thin-walled tubes is currently under development.107 This sleeve comprises hydraulically driven tubing arrays that contract and expand circumferentially when filled and emptied. As fluid enters the array of thin-walled polymer tubes connected side-to-side it transforms each tube from a flat

(deflated) to a circular (inflated) cross-section to effectively compress the epicardial surface in synchrony with ventricular ejection, ultimately leading to enclosed ventricular blood volume changes as high as 60%.107 This hydraulic DCCS device combined with the MEC technology introduced above could, in principle, allow for the development of a completely untethered, muscle-powered, non-blood-contacting VAD for long-term cardiac support.

27

1.4.2.2 Counterpulsation Extra-Aortic Balloon Pump

Another form of circulatory support for CHF patients that provides effective cardiac unloading and patient stabilization is displacement of blood from the aorta during the diastolic phase of the cardiac cycle. This technique is most often performed clinically using an IABP that is implanted and inflated inside of the descending aorta as previously described. This mechanical support augments diastolic pressure and coronary circulation via balloon inflation and reduces the resistance to systolic output via the presystolic deflation of the balloon.16,17,59,115 The biggest factor that prevents this technology from becoming a viable method of long-term support is the fact that it is often associated with thromboembolism with extended use due to its direct interaction with the blood stream. Therefore, an extra-aortic balloon pump (EABP) (Figure 8D) that wraps and compresses the external surface of the ascending aorta like the C-pulse device (Sunshine Heart

Inc., Eden Prairie, MN) may offer clinicians an alternative solution. The C-pulse counterpulsation

EABP was clinically tested and shown to significantly increase aortic peak diastolic pressure, cardiac output, and regional and coronary blood flow without touching the blood.103,108,116 In the context of long-term cardiac support it is worth mentioning that, like the soft robotic DCCS, this device also has the potential to be driven by muscle-powered actuation, which would allow for durations of use far beyond what is now possible with pneumatic actuation.

28

Figure 8. Biomimetic (A), minimally invasive (B), and muscle-powered (C) soft robotic direct cardiac compressive sleeves (DCCS) use copulsation and extra-aortic balloon pumps (EABP) (D) use counterpulsation techniques to enhance cardiac function without directly interacting with the bloodstream.107,108,113,114,117

1.4.2.3 Passive Periventricular Restraint

Passive periventricular restraint, which involves wrapping the entire epicardial surface with a sleeve-like prosthetic to provide circumferential diastolic support to the failing heart, is an approach that evolved from a surgical procedure known as cardiomyoplasty (CMP) in which the ventricles were wrapped with the latissimus dorsi muscle flap and stimulated to contract in synchrony with the systolic portion of the cardiac cycle. While CMP was effective in reducing wall stress, myocardial oxygen consumption and adverse ventricular remodeling, these benefits were found to persist in some patients even after the muscle flap stopped contracting, which suggested that these same effects might be produced via passive ventricular restraint alone. Toward that end, several passive prosthetic devices were developed to produce the same effects without resorting to the surgical complexities and post-surgical complications involved with LDM flap isolation and subsequent transplantation into the chest. Corcap (Acorn Cardiovascular, Saint Paul,

MN, USA) and Paracor HeartNet were two such devices that were designed to act more like the

29 passive LDM flap insofar as pressure was applied uniformly across the ventricular free walls. The

Acorn sleeve was a flexible, polyethylene-terephthalate mesh that was placed around the heart through a median sternotomy to provide end-diastolic support and reduced wall stress.118 The

Paracor device was formed from Nitinol wire mesh encased in silicone that exerted continuous elastic force on the heart throughout the cardiac cycle and could be deployed over the ventricles via an introducer sheath positioned over the cardiac apex through a mini-thoracotomy.119 Both devices were tested in limited clinical trials but, despite showing positive LV remodeling in a subset of patients with dilated cardiomyopathy, neither produced significant improvements in patient survival or quality of life and were subsequently taken off the market.

1.4.3 CADs in Summary

Key characteristics of the large and expanding family of cardiac assist devices developed in the past, used in the present and slated for the future are summarized in Table 1 below.

30

Table 1. Commonly used cardiac assist devices and their key characteristics (*IVC: inferior vena cava, FA: femoral artery, LA: left atrial, PA: pulmonary artery).120–131

31

1.4.4 Patient Management for Long-Term Treatment

Since 2001 when the Randomized Evaluation of Mechanical Assistance for the Treatment of Congestive Heart Failure (REMATCH) trial became the landmark study that established the benefits of implantable, pulsatile, and permanent VAD therapy in patients with late stage CHF, survival rates have improved to nearly 80% one-year after primary implantation due to a combination of refinements in patient selection strategy, surgical techniques, and peri-operative management.3,19,132,133 Even though the survival rate has gone up, late stage CHF patients still suffer from physical and psychological distress stemming from the lack of mobility and freedom.

As the 2018 ENDURANCE supplemental trial concluded, the ideal form of destination therapy should provide effective and comfortable long-term mechanical support with an emphasis not only on prolonging survival, but also reducing morbidity and improving overall quality-of-life.134

Considering that there are currently no practice guidelines for patient management, there is an urgent need for a more systematic and organized protocols for these patients. As the PREVENT trial highlights, more seamless, real-time communication between patients and caregivers is needed.17 Devices like CardioMEMS (St. Jude Medical, St. Paul, MN, USA) and HeartAssist-5

(ReliantHeart Inc., Houston, TX, USA) that have sensor and remote monitoring capabilities via cell phone or other portable devices were developed to meet this critical need.17,135 Overall, VADs with long-term reliability and low complication rates in combination with proper postoperative and follow-on care will together establish what may be considered a true destination therapy.

1.5 Project Relevance / Scientific Merit

Since its inception in the early 1960s, a remarkable amount of research and development has been performed in an effort to improve and expand the field of cardiac assist devices. As a

32 result, a wide array of cardiac assist technologies is available to clinicians today, each with their unique set of strengths and weaknesses, but all designed with one common goal in mind: to provide safe, reliable circulatory support however and whenever it is needed.

Of course, these challenges grow larger as rising levels and durations of support are required and it is important to continue to seek solutions that will free these patients from persistent physical risks and psychological distress. Toward that end, reducing device-related complications and eliminating the loss of freedom imposed by percutaneous tethers will be key factors in developing CADs that are truly suitable for long-term or permanent use. In addition, replacing current patient management practices with physician–patient interface systems that are more systematic, convenient, and effective will likely play a big role in improving the lives of CHF patients who must rely on life-sustaining devices for years on end. Fortunately, there is reason to expect that many of these improvements will be implemented in the not-too-distant future as steps to meet these challenges are currently being taken by several groups working to develop effective destination therapies with longer patient survival times and improved quality-of-life.

33

Chapter 2

Implantable Volume Amplification Mechanism (iVAM)

2.1 Foundation Technologies and Rationale

2.1.1 Harnessing Skeletal Muscle as an Internal Power Source

Given the persistent difficulties encountered with existing blood pump technologies used in the chronic setting, the promise of safe, reliable long-term circulatory support is likely to remain unfulfilled unless and until there is a fundamental change in the way we approach this problem.

The research by our group suggests an entirely new means to deal with longstanding problems created by drivelines and blood contacting surfaces. Our approach is to avoid them altogether by harnessing the body’s own endogenous energy stores (i.e., skeletal muscle) and applying this power to the external surface of the heart or ascending aorta.107,136,137

The primary enabling technology behind this work is an implantable muscle energy converter (MEC, Figure 9A) developed by the senior author (DRT) under the auspices of the

Whitaker Foundation and the National Institutes of Health.103 The MEC is an internal energy transfer mechanism that utilizes electrically stimulated latissimus dorsi muscle (LDM) as an endogenous power source and transmits this energy in hydraulic form. Through numerous iterative

34 design improvements implemented over several years, this device has been refined and shown to exhibit excellent anatomic fit, extreme mechanical durability and high energy transfer efficiency

(>90%) with the capacity to transmit more than 1.25 J of energy per actuation cycle.103,104,136,138

This compares favorably to the work generated by healthy left ventricles, which, when pumping

5.0 L/min against a mean pressure of 100 mmHg at 72 beats per minute, deliver 0.925 joules/stroke to the bloodstream. From an energetic perspective, this means that the MEC could, in principle, provide 100% systemic circulatory support if a target VAD could be made to deliver maximum sustainable MEC power to the bloodstream with an efficiency of 74% with every heartbeat.

Figure 9. Computer rendering of the MEC device shown in full (A) and 3/4 cross sectional (B) views. 1) Actuator Arm, 2) Rotary Cam, 3) Spring Bellows, 4) Piston, and 5) Outlet Port.

The LDM is especially well-suited for use as the MEC’s power source due to its large size, surgical accessibility, proximity to the thoracic cavity, and steady state work capacity sufficient for long-term cardiac support.139 Secure muscle-device fixation is achieved using an artificial tendon sewn into the humeral insertion of the LDM, which is then anchored to the actuator arm

(Figure 9B-1) of the MEC using a patented clamp-and-loop technique.103 LDM contractions are controlled by a programmable pacemaker-like device (cardiomyostimulator) that coordinates muscle activity with the cardiac cycle. As the actuator arm (Figure 9B-1) rotates upward in

35 response to LDM shortening, a rotary cam (Figure 9B-2) compresses a metallic spring bellows

(Figure 9B-3) and a piston (Figure 9B-4) located directly underneath, ejecting 5 mL of pressurized fluid through the outlet port (Figure 9B-5).

The potential advantages of this approach to long-term circulatory support are significant.

By efficiently translating stimulated contractile energy into hydraulic power, the MEC serves to both reduce the risk of infection across the skin and enhance patient quality-of-life by eliminating the need for external hardware components such as extracorporeal battery packs, transmission coils, and percutaneous drivelines. Moreover, muscle-powered VADs would, in principle, be far simpler to maintain and hence much less expensive in aggregate than traditional blood pumps used for destination therapy, thereby resulting in wider availability and reduced costs for healthcare providers.103,104,136,138

2.1.2 Extra-Aortic Counterpulsation

This novel method of converting contractile energy into hydraulic power attains its value when this energy is successfully delivered to the bloodstream. Since Kantrowitz et al. first introduced extra-aortic counterpulsation in the early 1950’s, the diastolic counterpulsation technique has been a commonly used cardio-therapeutic mechanism that assists left ventricular function and augments coronary blood flow by lowering pressure afterloads in the aorta while increasing coronary perfusion.103,104,138,140–142 More recently, this support mechanism has been typically accomplished with inflatable balloon pumps that displace blood from within the aorta during the filling phase of the cardiac cycle. However, unlike current intra-aortic balloon pumps

(IABPs) (Figure 10A) that displace blood from the inside of the vessel, extra-aortic balloon pumps

(EABP) (Figure 10B) squeeze the aorta from the outside. Although this method still retains the

36 risk of atheroembolism from atherosclerotic plaques due to the repeated compression of the ascending aorta, it largely precludes secondary complications associated with thromboembolism by avoiding contact with the bloodstream.143–147 In this context, the C-Pulse EABP (Sunshine Heart

Inc., Eden Prairie, MN) is an attractive target for musclepowered cardiac assist due to its low power requirements and positive results in early clinical trials.143,147 Results from a study performed in eight CHF patients over a 6-month period show that the C-Pulse balloon inflation

(20 mL) at the aortic root can effectively counterpulsate the heart and improve patient outcomes.148

Unfortunately, direct application of the MEC for C-Pulse actuation is not possible since the MEC was designed for high-pressure, low-volume (5 mL) energy transmission. Thus, to combine these two technologies into a single circulatory support system, an intermediate device is needed to increase effective MEC displacement volume to match the target inflation volume of the C-Pulse device. Toward that end, an implantable volume amplification mechanism (iVAM) was designed to boost MEC output volumes by a factor of four. This tether-free, non-bloodcontacting MEC- iVAM/EABP coupling will provide partial, but meaningful, circulatory support.

Figure 10. Unlike conventional intra-aortic copulsation balloon pump (A), extra-aortic counterpulsation balloon pump (B) wraps around the external surface of the ascending aorta and inflates and deflates in synchrony with ventricular diastole and systole (C), respectively, without touching the blood 148–150.

37

2.2 iVAM Design Elements

The following design elements and their attendant performance criteria were used to guide the course of iVAM development: A) volume amplification, B) anatomic fit, C) energy transfer efficiency, D) muscle force and speed requirements, E) work storage and delivery, F) material selection, and G) device durability. These seven key aspects of iVAM design are described below.

2.2.1 Volume Amplification

Volume amplification was accomplished using the area difference between two nested cylindrical spring bellows: an inner bellows that contains a partial vacuum within its area Ai and an outer bellows with an area Ao (Figure 11). The 5mL of pressurized fluid ejected from the MEC directly enters the iVAM, fills the annulus passage area (Aa) and extends the two bellows in unison.

The input volume is amplified as the bellows pushes down piston B, ejecting the output fluid towards the EABP. Fine tuning of these internal component dimensions was required to create a durable) and implantable (i.e., compact for comfortable fit) device that amplifies the 5 mL MEC output to meet the 20 mL EABP inflation requirement.

38

Figure 11. Free body diagram of the MEC-iVAM complex showing the force distribution profile created during the ejection phase of device actuation.

2.2.2 Anatomic Fit

The MEC-iVAM complex (Figure 12) was designed to sit across the ribcage with the upper portion of the MEC housing and sewing ring resting above the ribs, the lower portion of the MEC passing through the ribs, and the iVAM positioned below (i.e., completely within the chest cavity).

Optimum anatomic fit was accomplished by minimizing the device dimensions (10.3 cm x 12.9 cm x 6.2 cm) and stacking the two devices one against the other so that the iVAM portion of the combined device is able to sit comfortably against in inner lining of the chest wall with minimal lung displacement. This arrangement not only provides a low-profile transthoracic fit but also

39 reduces device weight and minimizes energy loss by shortening the fluid travel distance as explained in the following section.

Figure 12. The full (A) and 3/4 cross-sectional (B) views of the linearly stacked MEC-iVAM complex that ensures a comfortable fit above the ribcage as well as within the thoracic cavity.

2.2.3 Energy Transfer Efficiency

The MEC-iVAM complex was designed to minimize turbulence and pressure gradients

(∆P) throughout the fluid flow path. To accomplish this, parametric computational fluid dynamics

(CFD) simulations were run for five different MEC-iVAM interface opening designs (Table 2A) and ten different iVAM outlet port designs (Table 2B) using ANSYS Workbench (Canonsburg,

PA). After the simulations were evaluated for low pressure gradients, low turbulence kinetic energy (TKE) and high laminarity of the flow path, we chose a combination of the Interface

Opening Design #5 (Table 2A) and Outlet Port Design #10 (Table 2B), which features an enlarged flow path via eight interface openings positioned 30 degrees apart for low hydraulic resistance, a centered outlet port for minimal flow turbulence, and a linearly stacked configuration for short fluid travel distance.

40

Table 2. The pressure gradients (∆P), turbulence kinetic energies (TKE), and flow path profiles of five different MEC-iVAM interface opening designs (shown in 1/4 cross-sectional views) (A) and ten different iVAM outlet port designs (shown in 1/2 cross-sectional views) (B).

Design ∆P Max TKE TKE Profile Flow Path/Velocity Profile Iterations [psi] [ft2/s2] A) MEC-iVAM Interface Opening Designs

1 0.0401 1.140

2 0.0104 1.579

3 0.0197 1.356

4 0.0119 0.503

5 0.0238 0.473

B) iVAM Outlet Port Designs

1 0.031 3.846

2 0.041 1.967

3 0.210 4.451

41

4 0.209 4.924

5 0.210 5.148

6 0.660 3.938

7 0.531 5.221

8 0.552 2.342

9 0.121 2.152

10 0.162 2.006

To investigate the complete flow profile of the selected design combination, one-eighth of the MEC-iVAM complex 3D model was reconstructed for expedited flow analyses within the selected reduced volume. Boundary pressure and inlet and outlet flow rates were set to 1 atm and

19.9 mL/s, respectively, for water (ρH2O = 0.98 g/mL) operating at body temperature. Results show a laminar flow throughout the fluid path with the streamline velocity, density and trajectory (Figure

42

13A) and an even pressure drop profile (Figure 13B). The pressure drop across the entire fluid path was 0.0238 psi, which was small enough to be neglected with respect to overall energy loss calculations (< 0.2%).

Figure 13. CFD of a one-eighth section of the MEC-iVAM complex design that exhibits the most laminar flow (A) and even distribution of pressure gradient (B).

2.2.4 Muscle Force and Speed Requirements

The device is designed to operate at contractile force and velocity levels compatible with the functional capacity of fully conditioned human LDM, which at peak sustainable power production generates roughly 95 N force and shortens at a rate of 11 cm/s.151 To confirm the muscle’s ability to reliably power this device, actuation force requirements of the MEC-iVAM complex were calculated.

Device actuation begins with lifting the MEC actuator arm, which was designed with a 5N preload force to allow the actuator arm to overcome the passive resting tension of a fully-trained

LDM.103,138 Rapid rotation of the actuator arm (≤ 350 ms) is essential in order to complete inflation of the EABP during the first half of the diastolic period, but rapid return of the actuator arm to the home position is equally important since balloon deflation must be complete before the onset of

43 cardiac systole (Figure 10C).152,153 Hence, the distribution of actuator arm forces in both forward- and return-stroke directions is a critical design consideration. These forces were adjusted via manipulation of two dynamic components internal to the MEC-iVAM complex: 1) metal bellows spring constants and directions and 2) partial vacuum pressures.

Material, thickness, inner and outer diameter, number of diaphragms, and the contour of the convolutions determine the spring constant of the bellows. The iVAM bellows material and thickness were set to stainless steel and 0.0762 mm (0.003 in) for a high spring constant for appropriate force contribution. The number of bellows convolutions, stroke length, and inner and outer diameters were set to fourteen, 0.411 cm (0.162 in), 7.092 cm (2.792 in), and 8.189 cm (3.224 in), respectively, to create an acceptable bellows life span (i.e., 194,040,000 cycles). Both MEC and iVAM bellows were installed at partially compressed states and tuned to produce both preload return-stroke force (KMEC • ∆SMEC) and an opposing force in the forward-stroke direction (KiVAM

• ∆SiVAM). The partial vacuum spaces created within the MEC housing (PMEC_Vac • Am) and iVAM inner bellows (PMEC_Vac • Ai) add force in the return-stroke direction, which helps to store energy within the device during LDM contraction (FLDM) for a rapid EABP deflation and reset of the actuator arm between contractions. These force vectors were compiled into a free body diagram

(Figure 11) and a force balance equation (Eq. 1), which were used to determine that maximum actuation force.

∑Fy = −(Fcam) + (KMEC • SMEC) + (PMECVac • Am) − (KiVAM • SiVAM) + (PiVAM_Vac • Ai) − (PMEC • Aa) + (PiVAM • Ao) (1)

Equation (1) is a force balance equation of the MEC-iVAM complex at equilibrium; Fcam: cam force; KMEC: spring constant of the MEC bellows; SMEC: stroke length of the MEC bellows;

44

PMEC_Vac: pressure within the MEC vacuum space; Am: cross-sectional area of the MEC bellows;

KiVAM: spring constant of the iVAM bellows; SiVAM: stroke length of the iVAM bellows; PiVAM_Vac: pressure within the iVAM vacuum space; Ai: cross-sectional area of the inner iVAM bellows;

PMEC: hydraulic pressure of the 5mL fluid; Aa: annulus area between the inner and outer iVAM bellows; PiVAM: hydraulic pressure of the 20mL fluid; Ao: cross-sectional area of the outer iVAM bellows (Figure 11).

The initial MEC vacuum pressure (PMEC_Vac) and volume (VMEC_Vac) were set to 1 atm and

3 3 27.252 cm (1.663 in ), and the initial iVAM vacuum pressure (PiVAM_Vac) and volume (ViVAM_Vac) were set to -3.5 psig and 9.652 cm3 (0.589 in3), respectively (Figures 11 and 14). The actuation force requirement by the LDM (FLDM), which is one-eighth of the cam force (FCam) (Eq. 2), over the course of 90 degrees actuator arm rotation was computed and plotted (Figure 15A). The max actuation force against 1 atm was predicted to be 36.5 N by hand-calculations.

1 F = x F (2) LDM 8 cam

Figure 14. Schematics of the initial and final states of the MEC-iVAM complex.

45

Figure 15. A) Force required to lift the actuator arm over the course of forward stroke against 1 atm. B) Work distribution projection at each component of the MEC-iVAM complex (W1: work stored in the MEC bellows; W2: work stored in the MEC vacuum space; W3: work stored in the iVAM bellows; W4: work stored in the iVAM vacuum space; and W1~W4: the sum of W1, W2, W3, and W4) during the forward-stroke.

2.2.5 Work Storage and Delivery

The generatable LDM force applied over the muscle shortening length (d = 32mm) can be directly translated to the amount of energy producible by the muscle with each contraction. In this section, we compute how much of this energy will be distributed to where over the course of complete actuation cycles. One complete cycle is the sum of two phases: 1) the forward-stroke where the output fluid enters and inflates the EABP and 2) the return-stroke where the fluid exits and deflates the EABP. The total work (W) is how much contractile energy is required from the latissimus dorsi for proper EABP actuations against end-systolic pressure. A fraction of input work

(W) is distributed among four different ‘storage’ sites (W1: work stored in the MEC bellows; W2: work stored in the MEC vacuum space; W3: work stored in the iVAM bellows; and W4: work stored in the iVAM vacuum space) and the EABP (W5: work delivered to the balloon) during

46 forward-stroke (Eq. 3). During the return-stroke, the stored energies work to pull the fluid back into the MEC-iVAM system.

푊 = 푊1 + 푊2 + 푊3 + 푊4 + 푊5 (3)

Device actuation begins with lifting the actuator arm. As the actuator arm lifts, the rotary cam underneath pushes down piston A and compresses the MEC spring bellows (Figure 11).

During this forward-stroke, 0.0424 J of energy (W1) is stored in the MEC bellows (KMEC: spring constant; SMEC: stroke length of the MEC bellows over the course of forward-stroke) (Eq. 4).

푊1 = ∫(KMEC • SMEC) dSMEC (4)

As piston A lowers, the vacuum space in the MEC housing expands, creating a negative gauge pressure within (Figure 11). With increasing volume and decreasing pressure, 0.459 J of work (W2) is stored in the MEC vacuum space (PMEC_Vac: pressure; VMEC_Vac: volume of the MEC vacuum space over the course of forward-stroke) in a form of potential energy that is later used to pull the piston A back up during the return-stroke (Eq. 5).

푊2 = ∫(PMEC_Vac) dVMEC_Vac (5)

Conversely, the iVAM spring bellows contributes 0.126 J of energy (W3) to the forward- stroke as it expands from its initially installed compressed state as fluid is expelled from the iVAM

(Figure 11). Accordingly, the W3 stored in the iVAM bellows (KiVAM: spring constant; SiVAM: stroke

47 length of the iVAM bellows over the course of forward-stroke) values calculate to be negative (Eq.

6) in the overall energy balance equation.

푊3 = − ∫[KiVAM • SiVAM] dSiVAM (6)

The expanding iVAM bellows will push down piston B and increase the volume of the iVAM vacuum space, lowering the negative pressure within (Figure 11). This air pocket stores

0.731 J of potential energy (W4; PiVAM_Vac: pressure; ViVAM_Vac: volume of the iVAM vacuum space over the course of forward-stroke) which helps to retract the fluid from the balloon with muscle relaxation (Eq. 7).

푊4 = ∫(PiVAM_Vac) dViVAM_Vac (7)

The remainder (W5) is delivered to the counterpulsation EABP (PEABP: pressure; VEABP: volume of the EABP over the course of forward-stroke) to actuate it against patient’s aortic end- systolic pressure during forward-stroke (Eq. 8).

푊5 = ∫(PEABP) dVEABP (8)

Ideally, the sum of the energy storages and delivery (W1 + W2 + W3 + W4 + W5) should equal to the total work required by the LDM (W) because totality of the five distributed energy components would either contribute to aortic compression or EABP deflation and actuator arm return. In reality, however, the sum of these compartmental energies will not perfectly match up

48 with the total work put into the system owing to energy losses in the form of heat due to friction.

The exact amount of non-recoverable energy loss will be empirically measured via bench tests.

The theoretical energy distributions are compiled into a plot (Figure 15B).

2.2.6 Material Selection

A completed muscle-powered counterpulsation system will consist of the MEC, iVAM, connecting conduit, and EABP (Figure 16). Heat treatable Stainless Steel (AM 350) was chosen for the MEC-iVAM body build due to its superior biocompatibility and weldability.154,155 The excellent corrosion resistance and high fusibility of this material combine to form a robust weld between the bellows and housing, which is essential for device durability. The high spring constant of the Stainless Steel bellows (Kss = 14.9 N/mm) also add extra resistance and flex life to the device.154,155 Sterile deionized water is the energy transmission fluid of choice due to its high specific heat capacity, incompressibility, low density and low viscosity, which make the system less susceptible to temperature changes, turbulent flow, and energy losses over the course of device actuation. An implantable plastic material with high biocompatibility and flexibility such as

Polyurethane, Silicone, or PVC, designed to withstand pressurized fluid delivery over millions of cycles, would all be suitable for the tubing and balloon bodies. The tubing will be secured on both iVAM and EABP ends with implant-grade stainless steel band clips.

49

Figure 16. A) A 3D CAD model of a complete muscle-powered counterpulsation system (MEC- iVAM complex, connecting conduit, and EABP) isolated and B) implanted in a human thoracic cavity. The virtual model in reference to human anatomy validates comfortable fit and orientation (artificial tendon and muscle stimulator not shown).

2.2.7 Device Durability

The device, once implanted, is expected to function reliably for long-term, if not permanent, use. Engineering the bellows configurations most suitable for this extreme operational condition is the key to designing a durable device. Bellows height, width, effective area, convolution profile, and stroke length must all be carefully tuned to create appropriate volume amplification in a limited space while minimizing bellows flexion stress. The current bellows design amplifies fluid volume

50 displacement while incorporating the minimum bellows stroke lengths possible in this design space. The durability of the bellows was examined using ANSYS finite element analyses (FEA) to quantify flex life. The life expectancy of the MEC and iVAM bellows exceed 450 million and

194 million cycles (Figure 17), respectively, which surpass the number of cycles considered as

“infinite life” for AM 350 Stainless Steel (107 cycles) as per ASTM.154–156 Therefore, the current bellows design can be rated as “fatigue-free” for an infinite life span. Other internal components, including seals, camshaft and needle bearings, were also designed for extreme wear and biochemical resistance. The durability of the spring-energized polytetrafluoroethylene (PTFE) lip seals was tested in vitro using a cycling apparatus, which indicated no sign of wear throughout the test period. The only component that showed any amount of wear was the camshaft, which was very slight shown by electron microscopy.138

Figure 17. The minimum flex life of the iVAM bellows exceeded 194 million cycles, which translates to more than 6.2 years of operation at 60 BPM.

51

Chapter 3

Extra-Aortic Ventricular Assist Device (eVAD)

3.1 Prototyping and Assessment

The MEC-iVAM complex (Figure 18) was manufactured by Flexial Corporation

(Cookeville, TN), a high-aspect-ratio welded bellows and accumulators company that specializes in industrial level engineering and manufacturing for sophisticated and complex applications. The manufactured and assembled device was then bench tested to assess its viability as a long-term circulatory support device.

Figure 18. The MEC-iVAM complex manufactured and assembled.

52

3.1.1 Test Bench Setup

Dynamic testing of the counterpulsation system was first conducted in vitro to confirm proper system function and assess overall mechanical reliability prior to live animal trials. During this series of tests, muscular actuation was simulated via a programmable linear actuator (VLCT

45-0050-060R-AN, Animatics Inc., Elma, NY) (Figure 19-1) attached to a smart motor

(SM23216MH-EIP, Animatics Inc.) (Figures 19-2 and 20A) that features a microprocessor, servo amplifier, memory module, high capacity roller thrust bearing, and encoder. The actuator was attached to the actuator arm of the MEC (Figure 19-3) via a thin metal chain to simulate the pull of the LDM while allowing the actuator arm to reset without assistance from the linear actuator return stroke mechanism (as is the case with muscular actuation wherein the LDM actively shortens to empty the MEC pump and passively stretches as the device fills between contractions).

Motor speed and piston/MEC coupling dynamics were programmed via the Smart Motor Interface

(Animatics Inc.) (Figure 19-4) to replicate LDM actuation profiles, the primary components being a 32-mm draw over a 1/3-s ‘contraction’ period. Target speed was set to 96 mm/s at cycle rates of

60 bpm. A miniature in-line force transducer (Load Cell ELFS-T3E-250N, Entran, Edmonton,

Canada) (Figure 19-5) was placed between the linear actuator and the MEC actuator arm to monitor actuation dynamics and calculate total ‘contractile’ energy used to actuate the MEC- iVAM complex. The MEC-iVAM complex communicated with a counterpulsation EABP (C-

Pulse #93020, Sunshine Heart Inc., Tustin, CA) (Figures 19-6 and 20B) via a conduit identical to the internal driveline that will be used in subsequent implant trials. The EABP was then secured around a silicone replica of the ascending aorta (Model #1454, The Chamberlain Group, Oak

Brook, IL) (Figure 19-7) made to empty into a mock circulatory system adjusted to provide mean afterload pressures from 80 to 180 mmHg using a syringe and a clamp (Figure 19-8). The afterload

53 in the mock aorta was monitored real-time using a Harvard pressure transducer (Model #00153,

Harvard Apparatus, Holliston, MA) (Figure 19-9). These waveforms were amplified by an amplifier module (#PS-30A, Entran) (Figure 19-10), collected by a data acquisition system (USB-

6001, National Instruments, Austin, TX) (Figure 19-11), and recorded with LabView (National

Instruments) (Figure 19-4) to quantify MEC-iVAM/EABP coupling dynamics, determine energy transmission levels, and gauge the mechanical reliability of the actuation scheme.

Figure 19. Bench setup schematics for the muscle-powered counterpulsation VAD.

54

Figure 20. A) A linear actuator-smart motor system mimics the muscular contractile motion to actuate the MEC-iVAM complex. B) A C-pulse EABP device.

3.1.2 In Vitro Test Results for System Accuracy and Viability

A series of assessments were performed to check if the manufactured MEC-iVAM complex meets the following requirements: 1) volume amplification for proper EABP inflation, 2) sufficient balloon actuation force against hypertension, 3) minimal balloon actuation speed for tachycardia, and 4) muscle energy distribution for energy efficient actuation cycles.

3.1.2.1 iVAM Volume Amplification

The measured output fluid volume ejected from the manufactured MEC-iVAM assembly

(Figure 18) was 13.68 (± 0.41) mL, which was an amplification of roughly 2.73 times of the original MEC output volume. This is thought to be due, in large part, to manufacturing difficulties encountered during the fabrication of iVAM bellows convolutions (Figure 9B-3). Subsequent root- cause investigation revealed the pressure differentials in the iVAM vacuum space created during the inlet pressurization was large enough to have the edge welded bellows “bulge” or comply, changing the area ratio of the inner bellows and outer bellows creating the shortage in volume transfer. It is important to note, however, that although a volume amplification factor of 4x (20 mL) was originally targeted to match the larger size C-Pulse devices used to treat adult male heart

55 failure patients in early clinical trials, this initial 13.7 mL prototype was still able to produce clinically significant ascending aorta counterpulsation pressures in benchtop simulations.143,144,149

3.1.2.2 Force Competency for Hypertensive Cases

Peak actuation forces against 1 atm, normal blood pressure, and hypertension pressures were empirically measured with a miniature in-line load cell placed between a linear actuator and the MEC actuator arm. The 1 atm afterload was achieved when nothing was attached at the end of the MEC-iVAM outlet port. The afterloads of normal blood pressure and hypertension were achieved by pressurizing a mock silicone aorta with a syringe plug (Figure 21). Because the counterpulsation EABP is designed to inflate at the dicrotic notch (DN) of the aortic blood pressure waveform where the aortic valve closes, we set the DN pressure to 100 mmHg and 125 mmHg for normal and high blood pressure cases, respectively.156–159

Figure 21. The test bench setup in rest (A) and in actuation (B) with the MEC actuator arm pulled to a 90-degree angle against a pressurized mock silicone aorta.

The measured peak tensile force (FMEC-iVAM) against 1 atm was 35.7 (± 1.9) N (Figure 22A), which is comparable with the theoretical max actuation force of 36.5 N (∆ ≅ 2.15%) that was hand-

56 calculated in section 2.4. This test demonstrated that the tensile and contractile components (i.e., spring bellows and vacuum spaces) of the MEC-iVAM prototype performed precisely as designed.

The measured peak forces against normal and high blood pressures were 73.5 (± 1.5) N (Figure

22B) and 76.4 (± 4.7) N (Figure 22C). Considering an LDM of average mass can be trained to generate 95 N under peak sustainable power output conditions, these results (Figure 22D) confirm that this muscle is a viable power source for chronic cardiac counterpulsation in both normal and hypertensive patients.151

Figure 22. The peak MEC-iVAM actuation forces measured by a load cell placed between a linear actuator that was programmed to displace 32 mm at a speed of 96 mm/s and the MEC actuator arm that actuated against 1 atm (A), 100 mmHg (B), and 125 mmHg (C) were 35.7 (± 1.9) N, 73.5 (± 1.5) N, and 76.4 (± 4.7) N, respectively (D).

57

3.1.2.3 Speed Competency for Tachycardia

To synchronize balloon inflation to the dicrotic notch of the aortic blood pressure waveform where the aortic valve closes while achieving complete deflation prior to the beginning of the R-wave on the electrocardiogram where isovolumic ventricular contraction begins, rapid inflation and deflation of the balloon are essential.143,144,149 Since fully conditioned human LDM generates its peak sustainable power at a shortening rate of 11 cm/s, the shortest possible balloon inflation time is 0.291 seconds.151 The measured deflation time of a hydraulic balloon attached to the outlet port of the MEC-iVAM complex was 0.136 (± 0.0294) seconds, providing enough time to empty the balloon without interfering with ventricular ejection and reset the MEC-iVAM system in preparation for the next pump cycle. With this 0.427-second actuation period, the maximum cycle rate of the system (assuming a typical systolic period of 0.250 seconds) calculates to be roughly 89 bpm. This demonstrates that the actuation cycle rate of the system is applicable for both normal heart rates and cases of mild tachycardia.

3.1.2.4 Energy Distribution

Lastly, we assessed how LD muscle input energy was distributed during the actuation cycle.

The peak force required by the LDM to actuate the system (FLDM) was quantified by measuring the tensile force when the MEC actuator arm was fully lifted to 90 degrees. Actuation force was integrated over the muscle contraction distance (XLDM) to compute the input work. The amount of energy delivered to the bloodstream was found by integrating the measured aortic pressure (PAo) over blood volume displacement (VAo). The amount of energy stored in the MEC-iVAM complex was found by integrating the actuator arm lift force (FMEC-iVAM) over the arc length of the actuator arm (XMEC-iVAM) when nothing was attached to the outlet port of the MEC-iVAM complex at 1

58 atm. Ideally, all of input energy should be either delivered to the bloodstream or stored in the system. But in reality, there will always be some minor miscellaneous (Misc.) energy losses due to non-conservative friction forces at the bearing sites. The theoretical energy distribution is compiled to an equation below (Eq. 9).

∫(F퐿퐷푀) dX퐿퐷푀 = [ ∫(PAo) dVAo ] + [ ∫(F푀퐸퐶−푖푉퐴푀) dXMEC−iVAM ] + 푀푖푠푐. (9)

Against normal blood pressure (DN: 100 mmHg), it required 0.848 (± 0.032) J of energy to actuate the MEC-iVAM system, of which 0.221 (± 0.0042) J was transferred to the bloodstream during balloon inflation and 0.547 (± 0.0463) J was stored in the system for a rapid balloon deflation and return of the actuator arm to its original position. This indicates that 90.56 (± 6.58) % of the input energy was used towards EABP actuation. Against high blood pressure (DN: 125 mmHg), it required 0.866 (± 0.037) J of energy to actuate the system, of which 0.266 (± 0.0014) J was delivered to the EABP and 0.547 (± 0.0463) J was stored in the device. This means 93.88 (±

5.86) % of the input energy was used for the EABP actuation, making our muscle-powered counterpulsation VAD an energy efficient system. The remainder was most likely contributed to various processes such as the EABP pushing against elastic aortic wall, fluid flow resistance at turbulent regions near step-down connectors between the iVAM outlet port and the balloon hose, heat energy loss due to friction during fluid travel, and measurement errors either at force or pressure gauges.

59

3.1.3 Does Our Prototype Meet Design Criteria?

The following seven design criteria were used to guide the development of the muscle- powered counterpulsation system: 1) anatomic fit, 2) muscle force and speed requirements, 3) work storage and delivery, 4) energy transfer efficiency, 5) material selection, 6) volume amplification, and 7) device durability. Anatomic fit was achieved by linearly stacking two already-compact devices. The MEC-iVAM complex will sit across the upper left ribcage to ensure device stability and patient comfort. The sizes and inward pressures of both EABP and securing strap should also be carefully adjusted to meet individual patient anatomic requirements and account for conditions such as aortectasia and aortic aneurysm.160–162 The surgical approach and implant configuration plans are further described below. After completion of muscle force tests comparing peak actuation force requirement against high blood pressures and the LDM’s known peak sustainable force, and actuation speed requirements set by the minimum balloon inflation and deflation durations, the LDM was found to be an appropriate power source for cardiac assistance for hypertensive and tachycardiac patients. For muscle energy distribution analysis, results show that roughly 30.7 % of the input energy is transmitted to the bloodstream and 63.2 % is stored in the MEC-iVAM complex to power the return-strokes. Considering only about 6 % of the muscle energy is lost to processes like aortic wall compression and flow turbulence, the muscle-powered counterpulsation EABP can be considered an energy efficient system. Biocompatibility of the entire system is ensured by carefully selecting implant-safe materials such as medical-grade

Stainless Steel and Polyurethane for each part of the system. In regard to volume amplification, our first prototype amplifies the MEC output displacement volume by 2.73 times, yielding roughly

13.7 mL for EABP actuation. By refining manufacturing process of the iVAM, we will be able to produce 20 mL output volume for balloon actuation with the next prototype. Lastly, while the

60 durability of the MEC and iVAM bellows and other internal parts, including seals and bearings, met the criteria for an “infinite” life rating, the longevity of the flexible EABP should be confirmed via cycle testing on the bench.

3.1.4 Limitations

The shortage in volume amplification due to initial manufacturing limitations produces partial counterpulsation instead of the full 20 mL actuation originally targeted. Because this test was conducted with only our first prototype, this issue can be resolved when manufacturing next generation prototypes. Experimental limitations in force and pressure measurements may exist in forms of instrumental, observational, environmental, and/or theoretical errors. In an effort to minimize experimental errors, however, repeated measures design (N = 30) were used. Although the energy requirement for compressing the aortic wall in a pinching motion is expected to be miniscule, there is expected to be a small difference in wall resistances between an actual human aorta and the mock silicone aorta. This difference can be measured by repeating the test with a replaced human aorta in the future. Despite the fact that this muscle-powered counterpulsation

VAD is already an energy efficient system, we can further reduce energy waste by enlarging the balloon opening, thereby eliminating the need for step-down inline connectors between the iVAM outlet port and EABP. This will significantly lower flow turbulence and frictional heat loss.

Rigorous durability testing is essential for any destination therapy. Fatigue and cyclic loading of the soft polymeric EABP will be examined prior to preclinical testing.

61

3.2 Surgical Approach and Implant Configuration

The upper left ribcage is an ideal location for MEC-iVAM implantation due to its proximity to both the LDM insertion point and the ascending aorta. Comfortable implantation and secure fixation will be achieved by placing the device across a transthoracic window created by resection of a 6.5 cm portion of one rib. The MEC suture ring will be anchored to the adjacent ribs with wire suture while the bottom two-thirds of the device will fit across and within the chest wall as shown below (Figure 23). The outer (MEC) portion of the device will be oriented so that the direction of actuator arm rotation aligns with the direction of LDM shortening for maximum energy transfer efficiency. Inside the chest wall, the orientation of the outlet port will be arranged to stabilize the flow path and hence optimize fluid transfer efficiency between the MEC-iVAM complex and

EABP. Both left and right sides of the mid-sternal line are viable options for EABP placement.

Based on previous studies, the strength of fixation sites linking the device, muscle and chest wall is expected to improve and stabilize over time as fibrous tissue in-growth proceeds during the initial 2 to 4 weeks of device implantation.102,163,164

62

Figure 23. A complete eVAD system implanted in a human thoracic cavity.

3.3 Alternate Application

The MEC’s potential to drive pulsatile blood pumps extends to any form of hydraulic device designed to squeeze or otherwise manipulate the heart or aorta, preferably from the outside.

One such example is a device currently under development in our lab called a soft-robotic direct cardiac compression sleeve (DCCS).137,165 As previously explained in Chapter 1, the DCCS is a copulsation device that would operate in synchrony with left ventricular ejection. This alternative means to harness endogenous muscle power for cardiac assistance was 3D-printed using low modulus polymeric soft materials and developed in parallel with the extra-aortic counterpulsation system.

63

3.4 Conclusions

Whether the application is counterpulsation EABP or copulsation DCCS, the muscle- powered system serves to both reduce the risk of infection across the skin and enhance patient quality-of-life by eliminating the need for external hardware components such as extracorporeal battery packs, transmission coils, and percutaneous drivelines. Moreover, using muscle power to actuate non-blood-contacting pumps avoids thromboembolic events, and obviates the need for long-term antithrombotic therapies like anticoagulant drugs, antiplatelet agents, and routine surveillance.165 The muscle-powered VAD would, in principle, be a more attractive option for destination therapy as it would be simpler to maintain and hence less expensive in aggregate than traditional blood pumps, thereby resulting in wider availability and reduced costs for healthcare providers.

64

Chapter 4

Compressive Ventricular Assist Device (cVAD): Design

4.1 Soft Robotic Direct Cardiac Compression Sleeve (DCCS)

A non-blood-contacting direct cardiac compression sleeve (DCCS) is an alternate application of the muscle power for long-term circulatory support. We propose to use the MEC’s

1.25 J of contractile work to drive a soft-robotic DCCS that applies copulsation to both ventricles

(Figure 24). This approach is not new; in fact, numerous cardiac sleeves – including Mannequin

(Chase Medical, Richardson, TX) and Corcap (Acorn Cardiovascular, Saint Paul, MN) – have shown remarkable therapeutic benefits and are currently in clinical use.166,167 While it is unfortunate that these pneumatic DCCSs have not yet overcome the driveline infection problem, our completely implantable muscle-powered system will resolve this complication by avoiding the need to transmit energy across the skin altogether.

65

Figure 24. Computer rendering of a muscle-powered copulsation system implanted in a patient (stimulator not shown).

4.2 Sleeve Geometry and Concept

4.2.1 Geometry

Toward that end, we designed a soft-robotic DCCS to boost heart function without directly touching the bloodstream. The DCCS is a copulsation device that assists cardiac function by wrapping around and squeezing the external surface of the ventricles in synchrony with the contractile cycle of the heart. Considering a healthy heart has a ventricular ejection fraction (EF) ranging from 55% to 70% and anything less than that is diagnosed as mild (< 54%) to severe (<

35%) heart failure, our goal with DCCS support is to either bring patients’ EF back into the normal range or, in the case of dilated hearts, generate sufficient ventricular stroke volume to restore cardiac output to pre-disease levels.168

The DCCS currently under development in this lab uses the geometric advantage produced by an array of thin-walled tubes to compress the exterior of the ventricles during the systolic phase

66 of the cardiac cycle. These tubing arrays contract and expand circumferentially when filled and emptied. As fluid (either liquid or gas) enters the array of thin-walled polymer tubes connected side-to-side, it transforms them from an elliptical cross-sectional configuration to a circular one, causing the effective widths of the tubes as well as the radius of the DCCS device as a whole to decrease (Figure 25A). To be specific, when these active elements in the circular array are fully inflated, they form a perimeter of length nd, where n is the number of tubes in the array and d is the diameter of each individual tube (Figure 25B). Conversely, when fluid is removed, each tube collapses flat so that their effective width increases from their inflated diameter (d) to roughly one- half their inflated circumference (πd/2). Thus, each tubular element expands sideways by π/2 or

57% with deflation and the circumference of the circular array enlarges to πnd/2. These active elements can be arranged to form a soft-robotic hydraulic DCCS (Figure 25C) that covers and compresses the epicardial surface of the ventricles. The sleeve is composed of a hollow loop (Fig.

C-a) that evenly feeds fluid to contractile elements (Fig. C-b) connected with inextensible straps

(Fig. C-c) in between. By operating this hydraulic actuator in synchrony with ventricular (LV) ejection, we can, in principle, achieve enclosed ventricular blood volume changes as high as

60%.137

67

Figure 25. Thin-walled tubes are arranged in a circle and drawn toward the center during inflation (A). An array of contractile elements (B) forms a soft robotic cardiac compression sleeve (C). For (C), the sleeve is composed of a hollow loop (a) that evenly feeds fluid to contractile elements (b) connected with inextensible straps (c) in between.

To actuate these contractile elements around the ventricles in synchrony, a hollow hoop that evenly feeds fluid and distributes pressure into each tube is designed to be anchored around the base of the heart. The array of tubes extending from underneath the ring is in a half-prolate shape with the bottom quarter truncated that, when deflated, snuggly fits around the ventricles.

Inextensible straps were added between the tubes as a means to modulate overall DCCS displacement volume. With these design concepts in mind, computational modeling and static- structural finite element analysis (FEA) were done to examine the effects of fundamental design variables including tubing wall thickness (Th), Young’s modulus (E) of the tubing material, internal hydraulic pressure (P), and number of tubes (N) on overall function of the sleeve. The goal is to achieve an optimal sleeve design that experiences the least stress, but the most inward deformation possible.

68

4.2.2 Wall Thickness

When simulating the sleeve model with three different tubing wall thicknesses (0.7, 1.0, and 1.3mm), results from computational models show that a design with thinner tubing walls experiences less stress while deforming more (Figure 26A). Although it would therefore be ideal to make the walls as thin as possible as long as they endure the internal hydraulic pressure and force requirements, manufacturing constraints exist. In our experience, it is difficult to manufacture walls thinner than 1.0 mm while maintaining uniform thickness and tear resistance, even when considering a variety of manufacturing methods such as silicone molding and 3D printing.

4.2.3 Material Requirements

Data from FEA models used to simulate the sleeve with three different Young’s moduli (6,

9, and 12MPa) showed that devices made with lower durometer polymers will perform better

(Figure 26B). It is important to note that while high material flexibility is essential, it is also important that the sleeve exhibit low – if not close to zero – extensibility. Indeed, an ideal compressive sleeve, formed from a flexible biomaterial with low Young’s modulus and elongation ratio, would deform with minimal expansion, resulting in optimal inward compression and stress exertion. While limited material choices for current fabrication methods preclude our manufacturing the most ideal prototype at this time, new techniques are being developed to incorporate a wider array of biomaterials and mesh components to optimize device functionality, biocompatibility, and durability.

69

4.2.4 Pressure Requirements

Hydraulic pressure within the tubular elements of the device must be large enough to deform the sleeve and compress against the ventricles’ physiologic pressure without surpassing the yield stress of the tube walls. As expected, when simulating sleeve dynamics using three different hydraulic pressures (10, 20, and 30 psi) with other variables kept constant, the device with the highest internal hydraulic pressure exhibited the most deformation and stress (Figure 26C).

It is therefore essential that DCCS material properties be engineered to withstand maximum MEC output pressures to produce maximum compression while safely maintaining structural integrity through millions of duty cycles.

4.2.5 Number of Tubes

When simulating the sleeve model with three different tube and strap numbers (8, 12, and

16), the sleeve with a largest number of tubes resulted in the largest sleeve deformation (Figure

26D). This is not unexpected. In fact, fifty 0.15cm wide tubes in an 8cm tall half-prolate spheroid shape are computed to amplify hydraulic input volume by a factor of five.169 But because N’s greater than twelve have proven difficult to manufacture with prototyping technologies employed to date, our latest working prototype embodies a 12-tube design with twelve straps filling in the remaining sleeve circumference.

70

Figure 26. Bar graphs comparing the stress and geometric deformation experienced by FEA models with different A) wall thicknesses, B) materials, C) hydraulic pressures, and D) number of tubes.

4.2.6 Manufacturing

Our current soft-robotic DCCS design is an 8cm (w) x 8cm (l) x 6cm (h) sleeve composed of a circular manifold, 12 tubular actuators and 12 interstitial straps with a uniform wall thicknesses of 1.0 mm (Figure 27A). The width and height of the tubes were made to accommodate low- volume (5 cc) MEC actuation while the width of the straps were calculated to produce a close anatomic fit around a typical human heart with the sleeve deflated (note: sleeve size would be

71 made patient-specific in practice). This prototype was printed with a flexible PU polymer (E=7.7

MPa) using Formlabs 3D Printer (Figure 27B).

Figure 27. A) Computer rendering and B) 3D printed prototype of the current DCCS design.

4.2.7 Implantation

Again, the upper left ribcage is an ideal location for MEC implantation due to its proximity to both the LDM insertion point and the ventricles (Figure 24). Comfortable implantation and secure fixation will be achieved by placing the MEC across a transthoracic window created by resection of a 6.5 cm portion of one rib. The top ring of the DCCS will be sutured to the base of the heart with biocompatible adhesive such as BioGlue, used to further secure the device.

72

4.3 Sleeve Design Optimization via Finite Element Analysis

4.3.1 Materials and Methods

As previously mentioned, to boost cardiac output without risking thrombotic events associated with blood-contacting surfaces, there have been numerous developments of non-blood- contacting VADs that apply pressure to the exterior surface of the heart such as the Anstadt

Assistor Cup, DeBakey’s pneumatic compression cup, CorInnova’s minimally invasive DCCS and a biomimetic silicone sleeve that both compresses and twists the heart.15,109,114 Surprisingly, despite the large body of work centered around biventricular compression and the obvious differences in left/right ventricular anatomy and systemic/pulmonary afterload pressures, there has never been a systematic study examining the application of independent pressures over the left and right ventricles as a means to optimize cardiac output in heart failure patients. In this study section, our goal is to quantify pressure requirements by the DCC sleeve for clinical level improvements in both ventricles for effective long-term circulatory support. As a first step toward that goal, we examined the hemodynamic and tissue deformation effects produced by varying epicardial pressures (EPs) on the left and right ventricles in a passive finite element model of the human heart.

4.3.1.1 Biventricular Model Geometry

Using data from computerized tomography (CT) stereolithography scans (Figure 28A) of human hearts downloaded from open-source 3D CAD model depositories such as GrabCAD

(www. grabcad.com) and 3D CAD Browser (www.3dcadbrowser.com), a 3D echocardiogram model (Figure 28B) that was scanned at different directions and reconstructed by Insilicomed Inc.

(La Jolla, CA), and eight different literatures that reported average dimensions of human heart, a biventricular (BiV) model (11.6 cm (w) x 13.2 cm (l) x 10.5 cm (h)) (Figure 28C) with a prolate

73 left ventricle (LV) and a crescent-shaped right ventricle (RV) with wall thicknesses of 1.6 cm and

0.9 cm, respectively, was rendered using SolidWorks 3D design software (Dassault Systèmes

SolidWorks Corporation, Waltham, MA).170–177 The initial ventricular volumes prior to cardiac compression were set to normal end-diastolic values of 135 mL for the LV and 150 mL for the RV so that a stroke volume (SV) of 80 mL corresponds to roughly 60% left ventricular ejection (LVEF) and 53% right ventricular ejection fraction (RVEF).173 This computer-rendered BiV model was then converted to an initial graphics exchange specification (.IGES) format for cardiac compression finite element analysis (FEA) simulations.

Figure 28. Human biventricular models. Examples of human biventricular models reconstructed from a CT scan (A); an echocardiogram scan (B); and a computer rendering of a passive human biventricular model (C) designed with these reported data.

4.3.1.2 Constitutive Modeling of Passive Heart Tissue

A multiscale FEA software platform called Continuity, developed by the Cardiac

Mechanics Research Group at University of California San Diego (UCSD), was used to estimate the passive constitutive law of the myocardium employed in these simulations. Continuity and the derivative work, Continuity Pro (Insilicomed Inc.), combine high-order FEA of a nonlinear

74 constitutive law for the muscle fibers and a dynamic model of myocardial excitation-contraction coupling. The software was used to inflate a passive cylindrical model of cardiac tissue, which was extracted from a cohort of 13 patient-specific computational models based on medical imaging and other clinical measurements obtained from patients.178 A transversely isotropic constitutive law, adapted from previous work done by Guccione et al., was employed without any calcium activation.179 The original constitutive law was developed by using large deformation theory and previous stress-strain experiments on canine hearts to estimate parameters. The law employed here is a Fung type hyperelastic model with an exponential strain energy density function (Eq. 10).

퐶 푊 = (푒푄 − 1) (10) 2

in which C is a stress scaling coefficient, and Q is a quadratic function of the six normal strains and associated shear strains of a symmetric, Lagrangian finite strain tensor. Q has the following form (Eq. 11):

2 2 2 푄 = 푏2퐸푓푓 + 푏3( 퐸푐푐 + 퐸푟푟 + 2 ∗ 퐸푐푟 ∗ 퐸푟푐) + 푏4(2 ∗ 퐸푟푓 ∗ 퐸푓푟 + 2 ∗ 퐸푓푐 ∗ 퐸푐푓) (11)

The coordinate system for the strain tensor is the orthonormal basis of the local fiber coordinate system, with each fiber having a direction along its axis (f), a direction perpendicular to the fiber and directed along the surface of the heart (c), and a direction also perpendicular to the fiber directed transmurally (r). 퐸푓푓, 퐸푐푐, and 퐸푟푟 are the normal strains in the fiber, cross-fiber, and transmural or radial directions respectively. Likewise, 퐸푟푐, 퐸푟푓, and 퐸푓푐 are the shear strains. The variables 푏1 to 푏4 are absolute constants that scale the various strains. All the constants are

75 outlined in previous studies conducted with Continuity Pro but a few are reiterated here for convenience (Table 3).178

Table 3. Parameters used in the Passive Constitute Law of Myocardial Fibers.

In the cylinder, the fiber direction rotates helically from -37 degrees on the epicardium to

+83 degrees on the endocardium with respect to the circumferential direction. The inner wall of the cylinder was subjected to a linear increase in internal pressure up to 2 kPa. The strain of the middle layer of the myocardium was calculated to estimate the effect of the rotating fiber directions.

This strategy was selected as it would be almost impossible to recapitulate the fiber directions in

ANSYS and an aggregate estimation would be sufficient for a first order approximation. This stress-strain relation (Figure 29) was imported into ANSYS Mechanical and fit with a 3rd order

Yeoh hyperelastic equation for the strain energy density as depicted in an equation (Eq. 12). The

Yeoh form was chosen as it deals well with incompressible nonlinear elastic materials, its dependence on only the first invariant of the Cauchy-Green deformation tensor, and the ability to fit the relation well with polynomial shaped stress-strain curves.

3 푖 푊 = ∑푖=1 퐶푖(퐼1 − 3) (12)

76

Figure 29. Stress-strain curve of the heart tissue data overlay with Yeoh 3rd model.

Again, W is the resulting strain energy, 퐶푖 are constants determined empirically and 퐼1 is

2 2 2 the first invariant of the Cauchy-Green deformation tensor. 퐼1reduces to 휆1 + 휆2 + 휆3 , in which lambda1, lambda2 and lambda3 are the three principal stretches in which the principal axes coincide with the global coordinates of the cylinder, i.e., the circumferential, axial and radial coordinate directions. The expanded form of the Eq. 12 is presented below (Eq. 13).

2 3 푊 = 퐶1(퐼1 − 3) + 퐶2(퐼1 − 3) + 퐶3(퐼1 − 3) (13)

To estimate the parameters, the stress-strain curve was fit to the strain energy density function assuming the applied pressure was causing an equibiaxial extension of the tissue. The resulting parameters from this fit are: C1 = 2,734 Pa, C2 = 15,113 Pa and C3 = 89,498 Pa. The goodness of fit is illustrated in Figure 29 and shows good agreement with the calculated stress- strain curve. This constitutive heart tissue model was then evenly applied to the BiV model (Figure

77

28C) to imitate human heart tissue material properties without any myocardium layers or tissue fiber directions.

4.3.1.3 Boundary and Loading Conditions

Boundary and loading conditions were applied to the BiV model prior to running static structural FEA using ANSYS Workbench. The top surface of the model (Figure 30A) was set as a fixed support to mimic the natural constricted movement of the valvular plane of the heart. A small contact offset of 0.001m was added to the interior of the RV to prevent the inner surfaces (Figures

30B and 30C) from completely collapsing or penetrating each other. Afterload pressures were applied to the interior surfaces of the LV (Figure 30D) and RV (Figures 30B and 30C), and epicardial pressures (EPs) were applied to the exterior surfaces of the LV (Figure 30E) and RV

(Figure 30F).

Figure 30. Computer renderings of a biventricular model. The BiV model surface divided into six sections for proper boundary and loading conditions applications (A: Fixed top surface, B, C: RVP, D: LVP, E: LVEP, F: RVEP), and generated into a mesh model (G) for cavity extractions before (H) and after (I) deformations (J).

78

In this study, we investigated the range and relationship between applied external pressures and BiV model deformation under four different case scenarios: 1) uniform EP at zero afterload,

2) separate LVEP and RVEP at zero afterload, 3) uniform EP at end-systolic afterloads, and 4) separate LVEP and RVEP at end-systolic afterloads. For Cases 1 and 3, a uniform EP was evenly applied to the entire epicardial surface (Figures 30E and 30F) to simulate current direct cardiac compression methods. For Cases 2 and 4, LVEP and RVEP were applied to the epicardium of the

LV (Figure 30E) and RV (Figure 30F) separately to simulate discrete LV and RV compressions.

For zero afterload cases, both LV (Figure 30D) and RV (Figures 30B and 30C) afterloads were set to 0 mmHg to test the ventricular behavior solely dependent on its tissue material properties. For end-systolic cases, LV (Figure 30D) afterload was set to an aortic pressure (AoP) of 100 mmHg and RV (Figures 30B and 30C) afterload was set to a pulmonary artery pressure (PAP) of 25 mmHg to simulate direct cardiac compressions of a completely passive heart under normal circulatory pressure conditions.180–182

4.3.1.4 Mesh Cavity Extractions for SV and EF Analysis

To assess the effects of various loading conditions on the cardiac output of the BiV model, both initial and deformed mesh models (Figures 30G and 30J) were exported as stereolithography

(.STL) files and intersected for cavity extractions (Figures 30H and 30I) using SolidWorks. The stroke volumes were computed by subtracting the extracted final LV and RV cavity volumes from the end-diastolic LV and RV volumes before compression (Eq. 14). Ejection fractions were calculated by dividing the computed SVs by the initial end-diastolic volumes (Eq. 15).

푆푡푟표푘푒 푉표푙푢푚푒 (푆푉) = 퐸푛푑 퐷푖푎푠푡표푙푖푐 푉표푙푢푚푒 − 퐹푖푛푎푙 퐶푎푣푖푡푦 푉표푙푢푚푒 (14)

79

퐸푛푑 퐷푖푎푠푡표푙푖푐 푉표푙푢푚푒 − 퐹푖푛푎푙 퐶푎푣푖푡푦 푉표푙푢푚푒 퐸푗푒푐푡푖표푛 퐹푟푎푐푡푖표푛 (퐸퐹) = 푥 100 (15) 퐸푛푑 퐷푖푎푠푡표푙푖푐 푉표푙푢푚푒

4.3.2 Geometric Deformation and Cardiac Output Results

For Cases 1 and 3 (Figures 31A and 31B), we reported LV and RV deformations after uniform EPs were applied to the epicardial surfaces. For Cases 2 and 4 (Figures 31C and 31D), we investigated left and right EP combinations that induced comparable left and right SVs and EFs until the EF reached 50 ~ 55%, which is considered a normal value for both LV and RV.170,178 The

BiV model deformations of all four cases are illustrated in Figure 32.

Figure 31. Stroke Volumes and Ejection Fractions of the BiV Model after Four-case Cardiac Compressions. A: Stroke volumes (a) and ejection fractions (b) of Case 1 where uniform EPs were applied to both ventricles with zero afterload; B: Stroke volumes (a) and ejection fractions (b) of Case 3 where uniform EPs were applied to both ventricles with end-systolic afterloads; C: Cardiac outputs of Case 2 where separate EPs were applied to LV (a) and RV (b) with zero afterload; and D: Cardiac outputs of Case 4 where separate EPs were applied to LV (a) and RV (b) with end-systolic afterloads.

80

Figure 32. The BiV Model (A) Deformations after Four-case Cardiac Compressions. Case 1: A uniform EP of 5 mmHg (B), 10 mmHg (C), and 25 mmHg (D) applied to both ventricles with zero afterload; Case 2: Different combinations separate EPs (5 mmHg LVEP and 2 mmHg RVEP (E), 30 mmHg LVEP and 4 mmHg RVEP (F), 60 mmHg LVEP and 10 mmHg RVEP (G)) applied to both ventricles with zero afterload; Case 3: A uniform EP of 0 mmHg (H), 30 mmHg (I), and 60 mmHg (J) applied to LV and RV with end-systolic afterloads; and Case 4: Different combinations separate EPs (0 mmHg LVEP and 0 mmHg RVEP (K), 100 mmHg LVEP and 10 mmHg RVEP (L), 160 mmHg LVEP and 25 mmHg RVEP (M)) applied to LV and RV with end- systolic afterloads. Deformation units are in mm.

81

4.3.2.1 Uniform EP at Zero Afterload

For Case 1, a range of EPs were evenly applied to both ventricles with no afterloads inside to simulate a scenario where a symmetrical cardiac compression device pumps the ventricles of a completely passive, isolated heart. When the EPs ranging from 0 to 25 mmHg were applied uniformly across the epicardial surface, LVSV increased from 0 mL to 28.99 mL while RVSV increased from 0 mL to 101.93 mL (Figure 31A-a), inducing up to 21.27% LVEF and 67.68%

RVEF (Figure 31A-b). During compressions, the maximum deformation occurred at the midsection of the RV wall (Figures 32B - D), resulting in significantly more blood displacement from the RV than the LV (mean RVEF/LVEF ratio = 3.06).

4.3.2.2 Separate EPs at Zero Afterload

In Case 2, two individual EPs were applied to the LV and RV with no pressure afterload inside the heart. We looked for pressure combinations that induced similar LV and RV outputs

(difference less than 5%) until EFs exceeded 50%. The LVEP ranging from 0 to 200 mmHg induced LVSVs and LVEFs up to 75.77 mL and 55.59% respectively (Figure 31C-a), while

RVEPs ranging from 0 to 30 mmHg induced the RVSVs and RVEFs up to 78.48 mL and 52.11% respectively (Figure 31C-b). The maximum deformation during compression occurred at the apex of the heart (Figures 32E - G), which was due to shifts of the septal wall and the BiV model towards the RV in reaction to the much higher LVEP compared to the RVEP while maintaining similar LV and RV outputs. The relationship between LVEP and RVEP was close to linear with a mean

RVEP/LVEP ratio of 0.1473 (Figure 33A). These data indicate that it takes roughly 6.79 times the pressure to compress the LV than the RV for a completely passive heart model with zero afterload.

82

Figure 33. The RVEP-to-LVEP Ratio for Zero and End-systolic Afterloads. The relationship between applied LVEPs and RVEPs to induce the similar amounts of LVEFs and LVEFs (∆ <5%) for ventricles with zero (A) and end-systolic (B) afterloads.

4.3.2.3 Uniform EP at End-Systolic Afterloads

Although no cardiac compression device will normally have to support a completely passive heart since even the most severe heart failure cases exhibit 15 ~ 30% EFs, we tried compressing the passive BiV model against end-systolic afterloads (i.e., AoP = 100 mmHg, PAP

= 25 mmHg) as an indication of what might occur in a ‘worst-case’ scenario where ventricular fibrillation occurs during DCC support. Again, for Case 3, applications of uniform EPs to both ventricles represented cardiac compressions of a radially symmetrical pump. When EPs ranging from 0 to 60 mmHg were applied evenly to the epicardium, LVSV increased from -85.90 mL to -

63.60 mL and RVSV increased from -100.20 mL to 106.30 mL (Figure 31B-a). This corresponded to LVEF inflation from -63.02% to -46.66% and RVEF inflation from -66.53% to 70.58% (Figure

31B-b). Both ventricles started out in a bulged-out form with negative EFs because the afterload

83 was initially higher than the compression pressures outside. The LVEF remained in the negative range while RVEF passed +70% with 60 mmHg EP. The inward deformation of the RV wall was significantly more drastic compared to the LV wall (Figures 32H - J), which caused a dramatic difference in LV and RV cardiac outputs (mean RVEF/LVEF ratio = 9.34).

4.3.2.4 Separate EPs at End-Systolic Afterloads

When LVEP and RVEP were applied separately to look for EP combinations that induce comparable LV and RV outputs (difference less than 5%), LVEPs ranging from 0 to 340 mmHg induced LVSVs and LVEFs up to 73.86 mL and 54.19% respectively (Figure 31D-a), while

RVEPs ranging from 0 to 44 mmHg induced RVSVs and RVEFs up to 76.40 mL and 50.73% respectively (Figure 31D-b). Again, the maximum deformation occurred at the apex of the heart

(Figures 32K - M) due to the BiV model shift towards the RV to compensate for higher pressures over the LV. In this case the RVEP-to-LVEP ratio was roughly 0.1504 (Figure 33B), which means it requires about 6.64 times higher pressures on the LV than the RV to maintain similar LV and

RV outputs.

4.3.3 Stress, Strain, and Energy

Local maximum principal stresses and strains and total strain energy experienced by the

BiV model were studied for Case 4 since it is the simulation closest to physiologic cardiac compression. When maximum (σ1), middle (σ2) and minimum (σ3) principal stresses and maximum (ε1), middle (ε2) and minimum (ε3) principal strains were computed using ANSYS, both max stress and strain were initially located near the top surface of the BiV model due to the fixed boundary condition restricting movement (Figures 34A and B). As the BiV model was compressed

84 with higher EPs, however, the maximum strain locale transitioned to the middle of the RV wall

(Figure 34C). Also, because the BiV model is in a bulged-out state for LVEPs between 0 and 100 mmHg, the maximum principal stress initially dropped until the EPs exceeded the ventricular afterloads (i.e., LVP = 100 mmHg and RVP = 25 mmHg) (Figure 34D). Similarly, all three principal strains (ε1, ε2 and ε3) dropped before rising starting from 100 mmHg LVEP (Figure 34E).

Figure 34. Stress, Strain, and Energy Experienced by the BiV Model for Case 4. Local maximum (σ1), middle (σ2) and minimum (σ3) principal stresses and maximum (ε1), middle (ε2) and minimum (ε3) principal strains and total strain energy were solved using ANSYS Workbench for Case 4 where separate LVEP and RVEP were applied to the LV and RV loaded with end- systolic AoP and PAP, respectively. Principal stresses (A) initially dropped before rising due to the higher ventricular afterload compared to epicardial compressive pressure (D). Similarly, principal strains (B) dropped until the EP exceeded afterload pressures (E). Strain locale transitioned from the fixed top surface (B) to the middle of the RV wall (C) as EP increased. Total strain energy of the BiV model exhibited an exponential trend due to the hyperelastic material property of the model (F).

85

The total strain energy stored in the BiV model due to deformation was also computed by summing up probed energies for each mesh element using ANSYS Workbench (Figure 34F). The total strain energy experienced by the model was 0.483 J at 0 mmHg EPs and reached up to 2.88

J with 340 mmHg LVEP and 44 mmHg RVEP compressions. The strain energy is never zero because the BiV model experiences outward bulging even without any external compression

(Figure 32K), and it increases with an exponential trend due to Yeoh’s hyperelastic material properties of the model. The work in the LV and RV blood displacements at 54.2% LVEF and

50.7% RVEF were 0.985 J and 0.305 J, respectively, when hand-calculated by multiplying the afterloads with the corresponding volume displacements (Eqs. 16 and 17).

푊퐿푉_푏푙표표푑 = 퐸푛푑 푆푦푠푡표푙푖푐 퐴표푃 푥 ∆푉퐿푉_푏푙표표푑 (16)

푊푅푉_푏푙표표푑 = 퐸푛푑 푆푦푠푡표푙푖푐 푃퐴푃 푥 ∆푉푅푉_푏푙표표푑 (17)

These results indicate that 2.88 J are required for elastic deformation of the BiV model on top of 0.985 J and 0.305 J for LV and RV blood displacements in order to induce normal EFs. But because this simulation was modeled around several unavoidable assumptions, it does not necessarily mean the direct cardiac compression device needs to exert 4.17 J of energy for proper therapeutics (see Limitations section below).

4.3.4 Independent LV and RV Epicardial Pressures

Despite a number of investigations differentiating LV and RV epicardial compressions, including an inflatable “Heart Patch” device consisting of two separate patches placed on the left

86 and right ventricular free walls of a sheep heart and a soft robotic cardiac compression sleeve that selectively actuates LV and RV, no previous research studies have systematically examined the relationship, range, and effect of individual LV and RV epicardial compressions on a human cardiac model.178,179

As the gap between the LV and RV epicardial compression requirements was found to be not only present but also rather large (Table 4) due to the thinner wall thickness and lower afterload of the RV than those of the LV, it is safe to argue that separate applications of EPs on the LV and

RV are critical to maintaining comparable SVs in both ventricles.175,176 However, because no patient heart will be completely passive prior to the implantation of a DCC support device, the natural myocardial contraction and thickening dynamics of the beating heart will certainly reduce the amount of EPs required in practice. Therefore, the LVEP of 340 mmHg and RVEP of 44 mmHg requirements reported in this study represent an upper bound on the epicardial pressures needed to support a failing heart.

Table 4. A Summary Table of LVEP, RVEP and Total Strain Energy Requirements for LV and RV Ejection Fractions Higher than 50% for Both Case 2 and Case 4.

The difference between energy requirements for compressing against zero (Case 2) and systolic (Case 4) afterloads was another interesting finding of this study. Because epicardial

87 compression had to work against both elastic deformations of the BiV model and afterload pressures inside the ventricles, Case 4 required about twice as much energy as Case 2 (Table 4).

4.3.5 Limitations of the BiV Model Simulations

Although results from these preliminary simulations suggest that a cardiac compression device must deliver 2.88 J to deform the myocardium and an additional 1.29 J to move the blood against typical arterial afterloads in order to achieve a normal level of cardiac output in a passive heart (Case 4), it is important to point out that numerous assumptions and simplifications were made for this simulation study. Because these analyses were done on a simplified model with several underlying assumptions, these data do not necessarily indicate that an effective cardiac compression device must be designed to match these energy requirements.

The reasons are several. First of all, the geometry of the BiV model was designed based upon eight individual reference studies that report measurements of human heart dimensions. Due to substantial variations in patient size and measurement techniques, the BiV model used here is an idealized amalgamation and, like any such model, cannot profess to represent all patients of all stages of heart failure. Secondly, these simulations were done on a completely passive heart model that produces zero EF on its own, which, of necessity, will never be the case in clinical practice.

According to New York Heart Association (NYHA) Classifications, Class IV heart failure patients

(the most severe category) have an LVEF under 30%.183 Considering some level of active myocardial contraction is present even for Class IV end-stage heart failure patients, the energy requirement computed with a perfectly inactive model could be an overestimation. At the same time, due to the natural cardiac muscle thickening and tissue stiffening during systole, the energy requirement results may be an underestimation. Although this study was conducted with

88 considerable simplifications, there is no doubt that this pilot study is an incremental, and yet, crucial addition towards the development of an effective DCC device for long-term circulatory support. A more advanced simulation model that takes essential physiological phenomena – like myocardial stiffening and thickening during systole, competence of the atrioventricular and semilunar valves, variations in ventricular geometry, septal wall movements and hypertensive afterloads – into account will be a pivotal next step.

4.3.6 Future Steps

The DCC sleeve under development is a patient-specific device. Each device will be customized via fitting around a 3D reconstruction of a scanned patient heart. Repeating the cardiac compression simulations with a high-order biventricular model of a beating heart via multi-scale

FEA using Continuity Pro (Insilicomed, Inc.) and hemodynamic studies using fluid structure interaction analysis will provide a more accurate understanding of compression requirements and cardiac behaviors as well as ensure the viability of the device for various extreme cases such as hypertrophic cardiomyopathy.178 Once these requirements are further determined, a direct cardiac compression device that meets those criteria can be designed. One potential incarnation currently under development in our lab, a muscle-powered soft-robotic cardiac compression sleeve, is a completely tether-free ventricular assist system that compresses the epicardial surface of the heart and is powered by endogenous skeletal muscle. This device captures the contractile work of latissimus dorsi muscle and converts it into hydraulic power to actuate a pulsatile soft robotic pump, thereby enabling fully implantable circulatory support without the risk of driveline infections or thromboembolic events.107

89

As a preliminary exploration, this study provides fundamental insights and guidance towards the design of improved DCC devices for long-term circulatory support. The observations regarding resistance produced by both myocardial walls and pulmonary and arterial afterload pressures, and the relationship between LV and RV compression energy requirements, suggest that designing a two-part compression system that supports the LV and RV independently from each other will most effectively improve cardiac function.

90

Chapter 5

Compressive Ventricular Assist Device (cVAD): Prototype and

Assessment

5.1 Soft Material 3D Printing

5.1.1 3D Printing Soft Robotic Actuators

Soft robotic devices are becoming more and more prevalent in the biomedical field due to their high compliance and elasticity that allow safe device-tissue interactions. These advantages combined with additive manufacturing can fabricate functional soft robotic actuators that exhibit continuous and localized deformations with uniform force distributions.184,185 The fact that 3D printing technologies now allow fabrication of more complex designs with more sophisticated motion at a rapid prototyping speed and relatively low cost makes this approach an attractive option for building soft robotic devices for preclinical testing.

91

5.1.1.1 Additive Manufacturing of Flexible Materials

Additive manufacturing using 3D printers offers highly specific and reliable prototyping of sophisticated geometries with extremely low amounts of post-processing, risk of interfacial delamination, material waste, and cost.185,186 While these aspects make this cutting-edge technology an attractive choice for soft robotic fabrication, there also are limitations – especially when working with flexible materials. Soft, deformable materials in both filament and liquid resin forms are immensely challenging to print as per desired design specifications because they have a tendency to deform during the building process under their own weight.185,186 There are a number of ways to retain the desired design such as adding support structures and enhancing printing resolution by controlling the print speed and environment, but unfortunately, optimal 3D printing methods and settings are largely found by trial-and-error to this date.185,186

There are several different 3D printing methods currently available for this purpose including fused deposition modeling (FDM), direct ink writing (DIW), selective laser sintering

(SLS), stereolithography (SLA), and inkjet printing. Among these, FDM and SLA type printers are the most popular options due to the variety and depth of researches done on them. FDM printers build structures by melting and stacking a continuous filament of a thermoplastic material in the vertical, z-direction. They offer a handful of advantages like a higher range of printable geometries, build volume, cost efficiency, user-friendliness, and reliability. However, the technology is limited by low surface quality and resolution as they are restricted by the nozzle diameter.184,185 This low resolution often leads to leaking walls and it is a serious problem for soft robotics as airtightness is one of the most important requirements for a properly functional actuator. In SLA processing, liquid resin is selectively photopolymerized and cross-linked by local laser illumination. Because this technology can print thinner single layers, it allows for printing of thin, microscale features

92 and complex geometries with high resolutions.184,185 However, one major drawback of SLA printers is the lack of available materials with suitable mechanical properties for soft robotic applications.

5.1.1.2 3D Printable Soft Polymers

As limited as 3D printing methods and settings currently are, there are even fewer 3D printable soft polymer options suitable for building soft pneumatic or hydraulic actuators.184,185

The reason behind this is that proper device functionality largely depends on multiple key material properties, particularly Young’s modulus, ultimate elongation, and ultimate tensile strength.

Young’s modulus (E), or “tensile modulus,” is a mechanical property that defines the stiffness of a solid material. It is calculated by dividing the uniaxial stress over strain (Eq. 18) in the linear

(elastic) region of the stress-strain curve. The goal here is to build a soft robot with a material that has a Young’s modulus similar to that of biological soft tissue. Ultimate elongation (γ), or elongation at break, is the maximum elongation that a material can withstand before failure. It is calculated by dividing the change in gage length by the original gage length and multiplied by 100

(Eq. 19), where gage length is the original length of that portion of the specimen over which change in length is determined. Common soft robotic materials fail between 40% and 1,000% strain, but the requirement will vary depending on the intended use of the actuator. Ultimate tensile strength

(UTS) is the maximum amount of stress a material can handle before rupture. It is calculated by dividing the maximum load sustained by the specimen by the average original cross-sectional area

(Eq. 20). Required material characteristics will differ depending on the device’s target application, but generally speaking, soft robotics should have reasonable E, γ, and UTS for recoverable actuations without failure or fatigue.184,185

93

푈푛푖푎푥푖푎푙 푆푡푟푒푠푠 (18) 푌표푢푛푔′푠 푀표푑푢푙푢푠 (퐸) = 푆푡푟푎푖푛

훥 퐺푎푔푒 퐿푒푛푔푡ℎ (19) 푈푙푡푖푚푎푡푒 퐸푙표푛푔푎푡푖표푛 (훾) = 푋 100 푂푟푖푔푖푛푎푙 퐺푎푔푒 퐿푒푛푔푡ℎ

푀푎푥 퐿표푎푑 (20) 푈푙푡푖푚푎푡푒 푇푒푛푠푖푙푒 푆푡푟푒푛푔푡ℎ (푈푇푆) = 푂푟푖푔푖푛푎푙 퐶푟표푠푠 푆푒푐푡푖표푛푎푙 퐴푟푒푎

3D printing with soft polymers like thermoplastic polyurethanes (TPU), polydimethylsiloxanes (PDMS) silicones, polyimides, epoxies, polycarbonates, polyesters and hydrogels is becoming increasingly popular. But amongst them all, TPU and PDMS are predominantly used for soft robot fabrications because they require relatively less strict environmental control for stable device performance. TPU normally exist as solid polymers, usually in a filament form for FDM printers, at room temperature. Their biggest advantage is that they tend to smoothly pile up layer by layer without falling or bending due to their higher stiffness compared to other material types. But their toxicity and flammability concerns cannot be disregarded when developing medical devices, especially implants. PDMS are silicone rubbers with excellent elasticity and resilience. Because silicone elastomers are known for their chemical inertness, thermal resistance, biocompatibility, and low permeability, they are often the material of choice for applications in wearable sensors and soft actuators. However, silicone elastomers exist in a liquid state before crosslinking occurs and so cannot avoid the gravity-driven collapsing issue when printed using traditional techniques. Consequently, fine tuning of the material properties and printing environment are often required for silicone 3D printing.184–187

94

5.2 Evaluation of Five 3D Printed Sleeve Prototypes

The ideal material for our soft robotic DCCS is one that is extremely flexible and yet perfectly inextensible so that the device can easily actuate (rapidly inflate) while generating sufficient tensile force to compress the ventricles without stretching or ballooning of the contractile elements. (Note: This is especially important under pneumatic actuation conditions where the device is inflated to a target pressure rather than being filled with a given volume of fluid.) Here, we have selected five different 3D printable soft materials with shore durometers low enough to be a candidate for the soft robotic actuator’s body material and tested them in terms of 1) prototypability, 2) performance and functionality, and 3) implantability. The prototype must be printed precisely in the structure we designed and be functional without leakage. Once a printed prototype passed the first inspection point, its efficiency in volume amplification and energy was evaluated. Lastly, the material’s implantability in terms of adherence to the epicardial surface of the heart and overall biocompatibility was also examined.

5.2.1 Prototypability

A simplified version of the soft robotic DCCS was 3D printed with the selected five candidate soft polymers using printers and printing methods that were deemed most appropriate for each option. In this section, the material characteristics and printability were assessed to determine which of the five material options could potentially be used for our application.

5.2.1.1 Characterizations of Five Potential Material Options

Four TPU type materials, FormLabs Flexible resin (FormLabs, Somerville, MA), PolyFlex

(PolyMaker, Shanghai, China), FilaFlex (Recreus, Elda, Spain), and NinjaFlex (NinjaTek,

95

Manheim, PA), and one silicone type material, ChronoSil (AdvanSource Biomaterials,

Wilmington, MA), were chosen for testing due to their known low shore durometers and material properties that match our needs (Table 5). Two FDM type printers, UltiMaker 2+ (UltiMaker,

Geldermalsen, Netherlands) and LulzBot Taz 6 (LulzBot, Loveland, CO), and one SLA type printer, Form 2 (FormLabs), were used to print the chosen polymers.

Table 5. Print settings and material properties of the five soft polymers that were selected to be printed and tested.188–192

FormLabs Poly Flex FilaFlex NinjaFlex ChronoSil Flexible

AdvanSource Manufacturer FormLabs PolyMaker Recreus NinjaTek Biomaterials

Polymer Type TPE TPU TPU TPU Silicone

Printer Used Form 2 UltiMaker 2+ UltiMaker 2+ LulzBot Taz 6 LulzBot Taz 6

Printer Type SLA FDM FDM FDM FDM

Young’s Modulus (E) 3.88 MPa 9.4 MPa 3.62 MPa 12 MPa 16 MPa

Ultimate Elongation 80 ± 5 % 330.1 ± 14.9 % 665 % 660 % 550 ± 100 % (γ) Ultimate Tensile 8.1 ± 0.4 MPa 29 ± 2.8 MPa 42 MPa 26 MPa 35.4 ± 12.6 MPa Strength (UTS) Shore Hardness 80-85A 95A 82A 85A 80A

Price $199 / 1L $69.95 / 750g $43.00 / 450g $85 / 1kg $75 / 1lb Resolution (xy) 140 µm 0.25 - 0.8 mm 0.25 - 0.8 mm 0.6 mm 0.35 mm Resolution (z) 25 - 100 µm 20 - 600 µm 20 - 600 µm 0.05 - 0.4 mm 0.15 mm Print Speed ~ 3 layers/min 30 - 90 mm/s 20 - 40 mm/s 20 - 30 mm/s 9 mm/s Print Temperature 230 °C 220 - 235 °C 225 - 235 °C 230 °C 260 °C

Standard tensile tests were performed on the five selected material options to characterize and predict their behaviors. Nonrigid Type IV specimens (n = 3) of the five materials were 3D

96 printed with 100% infills (Figure 35A) following ASTM D638-14 Standard Test Method for

Tensile Properties of Plastics.193 A constant-rate-of-crosshead-movement type testing machine with hatched wooden grips for slip prevention was used to plot stress-strain curves of the five specimen samples (Figure 35B). For Formlabs Flexible, PolyFlex, FilaFlex, and NinjaFlex materials, the samples were imparted uniformly against a consistent pull at 60 mm/min using

Instron 4469 Universal Testing Machine (Instron, Norwood, MA) with 10kN load cell. ChronoSil specimen samples were imparted at 50 mm/min displacement rate using MTS Criterion 43

Universal Testing System with 5 kN load cell (MTS Systems Corportation, Eden Prairie, MN).

ChronoSil samples were manufactured and tested separately at Johns Hopkins University. Stress

-4 2 (σ) was computed by dividing the recorded load by the gauge area (AG = 1.50 x 10 m ), and strain (ε) was computed by dividing the recorded displacement by the initial length (Lo = 50 mm).

The Young’s modulus (E) of each material was computed by finding the slope of the initial section

(ε = 10%) of the stress-strain curve (SS-curve) because the trend was close to linear within this region. Young’s moduli found were 0.96 MPa for FormLabs, 5.28 MPa for PolyFlex, 1.92 MPa for FilaFlex, 2.40 MPa for NinjaFlex, and 14.52 MPa for ChronoSil. To characterize the elongation at break of each material, the specimen samples were pulled until failure. The ultimate elongations

(γ) measured were 63.74% for FormLabs, 660.9% for PolyFlex, 920.9% for FilaFlex, 953.3% for

NinjaFlex, and 410.3% for ChronoSil. The measured Young’s Moduli and maximum strain limit values of each material were compiled into a point plot for comparison.

We can use these stress-strain curves (Figure 35C) and E-γ plot (Figure 35D) to predict material behavior. As previously mentioned, we are looking for a material that is flexible, while tolerating a relatively low γ, because our application involves cyclical deformations without significant stretch. It is also important to note that the ideal material should have a linear elastic

97 response over a large strain, considering soft polymers generally exhibit plastic deformation after surpassing the linear elastic region and are unable to return to their initial state. The E-γ plot shows all five materials exhibit compatible Young’s moduli, while FormLabs and ChronoSil have lower

γ. Further investigations performed at the device level – including leak tests, adherence tests, and cyclic loading tests – will be required to ensure that the chosen material meets all durability and performance criteria for proper sleeve function.

98

Figure 35. A) Nonrigid Type IV specimen 3D printed with FormLabs Flexible, PolyFlex, FilaFlex, NinjaFlex, and ChronoSil (ordered from top to bottom) per ASTM guideline D638-14. B) A specimen sample is being tested with Instron 4469 Universal Testing Machine. C) Stress-strain curves of the five soft polymers chosen to be tested for a potential DCCS prototyping material. D) Young’s modulus and ultimate elongation plotted for the five materials.

99

5.2.1.2 Print Settings

For this preliminary material selection stage, we simplified the cardiac sleeve design into a 12-tube cylindrical sleeve actuator (Figure 36A) due to the complexities involved in printing radially curved surfaces in most of currently available 3D printers. This simplified cylindrical design with 6 cm diameter, 5.5 cm height, and 0.6 mm wall thicknesses was drawn with

SolidWorks and printed with the five chosen soft polymers with 100% infills and low print speeds to avoid material collapse. FormLabs Flexible resin was printed with FormLabs Form 2 at a printing speed of roughly 3 layers per minute and a printing temperature of 230°C. Both PolyFlex and FilaFlex were printed with UltiMaker 2+ at a printing speed of 30 mm/s and a printing temperature of 230°C. NinjaFlex was printed with a conventional LulzBot Taz 6 printer at speed and temperature of 30 mm/s and 230°C, respectively. ChronoSil sample was printed with a customized LulzBot Taz 6 at a 9 mm/s printing speed and 260°C printing temperature. It was also important to maintain a stable room temperature and humidity levels to near 20%.

5.2.1.3 Two out of the Five Printed Prototypes Pass

Despite carefully selected print settings and strict environmental controls, the quality of the

3D printed prototypes varied significantly due to the differences in each polymer’s chemical components, material properties, and adherability between printed layers in the z-direction as well as the printers’ nozzle sizes and resolutions in xy-directions. Under visual inspections, PolyFlex

(Figure 36C), FilaFlex (Figure 36D), and NinjaFlex (Figure 36E) prototypes exhibited more flaky, unfinished looks compared to FormLabs (Figure 36B) and ChronoSil (Figure 36F) prototypes.

When cross-sections perpendicular to the base of the five prototypes were imaged using an optical

CT scanner (Ganymede FD-OCT, ThorLabs, Newton, NJ), PolyFlex (Figure 36G), FilaFlex

100

(Figure 36H), and NinjaFlex (Figure 36I) scans showed uneven surfaces with grooves ranging from 0.1 mm to 0.5 mm, while FormLabs (Figure 36J) and Chronosil (Figure 36K) scans exhibited smoother surfaces with grooves less than 0.07 mm. These OCT images confirmed the visual inspection results.

Leakage testing is one of the most important (and most obvious) quality inspections that these prototypes must pass to qualify for use as a functional actuator. The soft robotic DCCS is essentially a hydraulic pump, where watertight seals all around the device are an absolute requirement. As mentioned above, there were severe variations in prototype qualities from one material to another even when print settings were optimized for high resolution prints (e.g., slow print speeding and high infill rate). In fact, only two prototypes – printed with FormLabs Flexible

(Figure 36B) and ChronoSil (Figure 36F) – passed the leak test. The other three did not exhibit watertight seals and had multiple leakage spots mainly due to adherence failures between print layers in the z-direction and fillet failures near sharp edges of the inlet port. Because three out of the five failed this qualitative leak test, only the two that passed this checkpoint were qualified to move on to further investigations required for the functional device development process.

101

Figure 36. A 12-tube cylindrical sleeve actuator was designed (W 6cm x L 6cm x H 5.5cm) with SolidWorks (A) and printed with FormLabs Flexible (B), PolyFlex (C), FilaFlex (D), NinjaFlex (E), and ChronoSil (F) for prototyping a preliminary soft robotic DCCS for a material selection process. Cross-sections of PolyFlex (G), FilaFlex (H), NinjaFlex (I), FormLabs Flexible (J), and Chronosil (K) prints were imaged with a ThorLabs OCT scanner.

102

5.2.2 Device Performance and Functionality

An ideal soft robotic DCCS should be able to bring the diseased ventricular EF back to a healthy level with limited amounts of hydraulic and energy inputs. In this section, volume amplification ratios and work requirements of the two soft robotic actuator prototypes that passed the leak test were assessed.

5.2.2.1 In Vitro Test Setup and Methods

By displacing ventricular blood volume that is multiples of the input volume, the device is essentially performing volume amplification. The volume amplification ratio is calculated by dividing the measured volume displacement by the amount of fluid injected into the soft robotic sleeve (Eq. 21). A higher volume amplification ratio signifies a more mechanically efficient device, and we expect the volume amplification ratios of the FormLabs Flexible and ChronoSil prints to be close to identical to each other due to their identical geometric structures. The device should also be efficient in terms of energy transmittance. To determine the amount of work required to actuate the sleeve, hydraulic pressure (Phyd) is measured as the input fluid was injected to the sleeve.

The amount of work (W) required to actuate the FormLabs Flexible and ChronoSil prints is then computed by integrating the Phyd over the amount of blood volume displaced (Vblood) measured on the bench (Eq. 22). Less amount of work required to actuate a device signifies a more energy efficient device. We expect the ChronoSil prototype to require a slightly larger amount of work than the FormLabs Flexible prototype to displace the same of amount of blood volume due to

Chronosil’s higher Young’s Modulus.

퐵푙표표푑 푉표푙푢푚푒 퐷푖푠푝푙푎푐푒푚푒푛푡 푉표푙푢푚푒 퐴푚푝푙푖푓푖푐푎푡푖표푛 푅푎푡푖표 = (21) 퐼푛푝푢푡 퐻푦푑푟푎푢푙푖푐 푉표푙푢푚푒

103

푊 = ∫(Phyd) dVblood (22)

To empirically measure the volume amplification ratios and calculate work requirements, a test bench was set up with a syringe (Figure 37A-a) to inject fluid into the actuator (Figure 37A- b) and a graduated cylinder (Figure 37A-c) to measure the blood volume displacement from a water-filled plastic bag (Figure 37A-d) that mimics a ventricular chamber in open air (1 atm). The hydraulic pressure inside of the actuator was monitored in real time using a Harvard pressure transducer (Model #00153, Harvard Apparatus, Holliston, MA) (Figure 37A-e). These waveforms were amplified by an amplifier module (#PS-30A, Entran) (Figure 37A-f), collected by a data acquisition system (USB-6001, National Instruments, Austin, TX) (Figure 37A-g), and recorded with LabView (National Instruments) (Figure 37A-h) to quantify energy transmission levels and determine the mechanical reliability of the actuation scheme.

104

Figure 37. Schematics (A – a)syringe, b) preliminary DCCS prototype, c) graduated cylinder, d) water-filled plastic bag, e) pressure transducer, f) signal amplification module, g) data acquisition system, and h) LabVIEW for data collection) and actual setups of the test bench with a FormLabs Flexible prototype (B) and a ChronoSil prototype (C) for volume amplification and work requirement assessments. Blood volume displacement (D) and work requirement (E) measurements were plotted.

105

5.2.2.2 Volume Amplification and Work Requirement

When 25 mL of input hydraulic fluid was injected into each of the two soft robotic actuators, blood volumes of 18.83 (± 1.04) mL and 20.00 (± 1.00) mL were displaced by the FormLabs

Flexible prototype (Figure 37B) and ChronoSil prototype (Figure 37C), respectively. The amounts of blood volume displaced by the two prototypes were close to each other as we expected (Figure

37D). The small difference present are potentially caused by 1) a marginal geometric difference

(e.g., wall thicknesses) in two prototypes due to different 3D printing methods used with different nozzle sizes and resolutions and 2) a systematic error due to micro air bubbles trapped in the actuators. The fact that neither prototype amplified the volume can be altered by modifying the device design once an adequate material is selected. The work requirements for device inflations were 0.415 (± 0.038) J for FormLabs Flexible and 0.706 (± 0.0068) J for ChronoSil (Figure 37E).

As we expected, again, the ChronoSil prototype required more energy than the FormLabs prototype to inflate the same amount. However, because both measurements fall under the range of generatable energy per stroke for our application (i.e., 1.25 J per actuation cycle), both prototypes pass as viable options for this test.194

5.2.3 Device Implantability

Safe and secure implantation is essential for any implantable device, and our soft robotic

DCCS is no exception. In this section, device implantability of the two passed 3D printed prototypes was assessed in terms of 1) adherability to the epicardium, 2) durability for repeated and prolonged actuation cycles, and 3) biocompatibility of the materials.

106

5.2.3.1 Lap Shear Test on Lamb Heart Epicardium

Secure attachment between the soft robotic sleeve and epicardium is essential for effective cardiac compressions. To that end, adhesion energies created between the two passed 3D printed materials and the epicardial surface of lamb hearts (Figure 38A) using two types of cyanoacrylate bio-adhesives were explored in this study. Following ASTM D1002-10, rectangular FormLabs

Flexible and ChronoSil samples were 3D printed and adhered to lamb heart epicardium strips

(Figure 38B) using Vetbond (3M, Maplewood, MN) and GLUture (Zoetis, Parsippany-Troy Hills,

NJ) to determine lap shear stresses.195 We selected Vetbond (99.9% N-butyl cyanoacrylate) and

GLUture (60% 2-octyl cyanoacrylate and 40% N-butyl cyanoacrylate) for their known biocompatibility and strong adhesion to tissues.196,197 The overlapping adhesion area was controlled to 20 mm x 25 mm and cured at room temperature for both FormLabs (Figure 38C) and

ChronoSil (Figure 38D) samples. An Instron 4400 Universal Testing Machine with a 30 kN load cell and mechanically tightened grips was used to measure tensile force against displacement of the crosshead. The crosshead was moved at a rate of 0.1 mm/sec until adhesive failure occurred.

The recorded load up to peak index (Figure 38E) was integrated over the crosshead displacement and divided by the initial overlapping adhesion area to obtain the adhesion energy per area [J/m2].

107

Figure 38. Lamb heart (A) epicardium strips (B) were adhered to FormLabs Flexible (C) and ChronoSil (D) samples for lap shear testing. An Instron 4400 Universal Testing Machine was used to measure the tensile force against crosshead displacement (E).

Based on the initial lap shear tests investigating the four different material-adhesive- epicardium combinations, the adhesion energies determined were 188.31 J/m2, 406.91 J/m2,

317.57 J/m2, and 444.79 J/m2 for FormLabs-Vetbond, FormLabs-GLUture, ChronoSil-Vetbond, and ChronoSil-GLUture combinations, accordingly (Table 6). GLUture exhibited stronger adhesions than Vetbond while ChronoSil showed higher adhesion energy than FormLabs.

Consequently, the ChronoSil-GLUture sample proved to be the strongest adhered combination.

108

Table 6. Adhesion energies generated from lap shear testing of FormLabs-Vetbond, Formlabs- GLUture, ChronoSil-Vetbond, and ChronoSil-GLUture samples.

Adhesion Energy per m2 [J/m2] Adhesive FormLabs ChronoSil

Vetbond 188.31 317.57 GLUture 406.91 444.79

When 11 N-cm and 20 N-cm of torque were applied to the porcine and silicone heart, the adhesion energy requirements found were between 53.9 J/m2 and 98.04 J/m2, suggesting the true adhesion energy of the system lies somewhere between these lower and upper bounds.198 Because the measured adhesion energies of all four material-adhesive-epicardium combinations (Table 6) exceed the upper bound by a significant margin, it is safe to conclude that any of the four combinations could withstand the shearing loads typical of this application and be qualified as good candidates for DCCS attachment.

5.2.3.2 High Frequency Cyclical Loading Test

The soft robotic DCCS, once implanted, is expected to function for long-term, if not permanent, use. Therefore, durability assessment of the sleeve material for repeated and prolonged actuation cycles is one of the most important evaluation procedures to complete. Toward that end,

DCCS actuation was first simulated using ANSYS Workbench finite element analysis (FEA) software to find the magnitude of the maximum local stress. These simulations showed that the maximum local stress (σmax) a sleeve material must tolerate for successful sleeve actuation without fatigue was 0.404 MPa near creasing hinges (Figure 39A). This means a sleeve material must be able to endure a stress of 0.404 MPa or higher at all times during cyclical testing to be accepted

109 for our application. To be conservative, however, the dynamic mechanical behavior and material longevity were evaluated under stress levels that were three times larger than the maximum local stress. Considering a gauge stress with a safety factor of three (σg = 1.212 MPa), the FormLabs

Flexible material was rejected as the material ruptured at a loading stress of 0.458 MPa (Figure

35C). Therefore, only ChronoSil was assessed for long-term durability in this section.

For high frequency cyclical load testing, dynamic mechanical analysis (DMA) was performed on a 3D printed ChronoSil specimen sample (dimensions: 2.15 mm width x 0.54 mm thickness x 4.90 mm loading gap; 100% infill) (Figure 39B) using RSA-G2 Solids Analyzer and

TRIOS Software v5.0.0 (TA Instrument, New Castle, DE) as per ASTM guideline D5026.199 The loading strain (εL) was set to 10% as it corresponds to our 1.212 MPa gauge stress according to the SS-curve of ChronoSil (Figure 35C). As for frequency, the sleeve inflation and deflation should complete within the ventricular systolic phase and the first 1/3 of ventricular diastolic phase, respectively, to deliver proper cardiac compression without interfering with ventricular filling between beats. Considering tachycardiac cases (i.e., heart rate of 120 BPM), the ideal loading frequency was determined to be 2 Hz. However, a frequency sweep test performed within a loading frequency range of 0.1 ~ 40 Hz demonstrated increased storage modulus (E’) and loss modulus

(E”) at higher frequencies (Figure 39C), which means a higher loading frequency applies a harsher condition to the sample. In order to shorten the experiment duration time as well as to create a more conservative test condition, the loading frequency (fL) was set to 10 Hz. Under these settings, uniaxial stresses experienced by the ChronoSil sample were measured for 1 million cycles (Figure

39D).

ChronoSil survived 1 million cycles without failure (Figure 39D). The stress experienced by the sample decreased from 1.40 MPa to 0.75 MPa over the course of 1M cycles due to multiple

110 factors such as residual stress caused by the high fL setting, internal changes in material particle alignments, polymeric viscoelasticity, and stress relaxation effect. Despite this material saturation observed during repeated loadings, the stress experienced by ChronoSil eventually plateaued at around 0.75 MPa, exceeding σmax at all times (Figure 39D).

Tensile tests on both pristine and post-1M cycle samples were then performed to compare their before and after mechanical properties (Figure 39E). Both before (Ebefore = 12.81 MPa) and after (Eafter = 8.15 MPa) Young’s Moduli remained within a viable range (Figure 35D). On top of that, both samples showed strain-softening before 100% strain and strain-hardening after 150% strain (Figure 39E). The similarity in these hardening trends signifies that ChronoSil retains its mechanical properties without experiencing plastic deformation or permanent rupture over the 1M- cycle actuation test despite harsh loading conditions like 10% strain and 10 Hz frequency.

111

Figure 39. A) Stress distribution on inflated DCCS tubes simulated by ANSYS Workbench. B) A ChronoSil sample being tested with RSA-G2 Solids Analyzer. C) Storage (E’) and loss (E”) moduli of ChronoSil for loading frequencies from 0.1 Hz to 40 Hz. D) Uniaxial stress experienced by the ChronoSil sample at εL of 10% and fL of 10 Hz for 1M cycles. E) A representative stress- strain curve of ChronoSil before and after the 1M cyclic loading test.

112

5.2.3.3 Biocompatibility

The soft robotic DCCS will be a completely implantable device, and therefore, biocompatibility is a fundamental requirement to meet prior to finalizing the material selection.

Despite the fact that FormLabs Flexible resin a reasonably good for prototyping soft-touch devices and adding ergonomic features to them, it has not yet been tested for any biocompatibility or medical-grade implantability.200 Conversely, ChronoSil is a family of polycarbonate-based silicone elastomers that exhibits not only superior pressure resistance, tensile-strength, and chemical resistance, but also biocompatibility.192 As for the adhesives, both Vetbond and GLUture are considered biocompatible for human tissue. Other medical-grade flexible bio-glues such as tough adhesives and Therapeutic Acrylates could also be potential candidates for the DCCS attachment.201,202

5.3 Ex Vivo Implementation on Fetal Bovine Hearts

Although the ChronoSil prototype is only a preliminary miniature version, we demonstrated its viability as a DCCS on passive fetal bovine hearts in this section. Three 2nd trimester fetal bovine hearts (Figure 40A) with average widths of 55.7 ± 5.9 mm, 48.3 ± 6.3 mm, and 28.7 ± 5.0 mm at the valvular plane, mid-plane, and the 25th percentile plane from the apex, respectively, were first used to examine the fitting of the miniature sleeve. Then, the sleeve was anchored to one of them to apply cardiac compression. A simple test bench was set up with a syringe to inject fluid into the DCCS and a graduated cylinder to measure the blood volume displacement. The inlet ring of the DCCS was aligned to the valvular plane of the heart to most effectively compress the ventricles. Left ventricular end-diastolic volume (LVEDV) was measured by slowly injecting water into the aorta of an empty heart until full. Left ventricular ejection

113 volume (LVEV) was found by measuring the volume change in the LV during repeated sleeve deflations (Figure 40B) and inflations (Figure 40C). Left ventricular ejection fraction (LVEF) was then calculated by dividing LVEV by LVEDV. The measured LVEDV was 5.4 mL and LVEV was 3.0 mL, resulting the LVEF of 55.6%. Despite numerous limitations of the current DCCS prototype, the possibility of cardiac assistance via a soft robotic cardiac compression mechanism was successfully demonstrated.

114

Figure 40. Three 2nd trimester fetal bovine hearts (A) were used to demonstrate cardiac compression induced by ChronoSil sleeve deflations (B) and inflations (C).

115

5.4 Conclusion and Discussions

Table 7. A summary table of assessments of the five selected 3D printable flexible polymers for a soft robotic DCCS material.

Material Implementation Leak Volume Energy Cyclical Materials Characteristics Adherence Biocompatibility on Fetal Test Amplification Requirement Loading Test E [MPa] γ [%] Bovine Hearts

PolyFlex 5.28 660.9 Fail ------

FilaFlex 1.92 920.9 Fail ------

NinjaFlex 2.40 953.3 Fail ------

FormLabs 0.96 63.7 Pass Pass Pass Pass Fail Fail - Flexible

ChronoSil 14.52 410.3 Pass Pass Pass Pass Pass Pass Pass

Five 3D printable flexible polymers were tested in terms of prototypability, performance

and functionality, and implantability for a soft robotic DCCS body material, the results of which

are summarized in Table 7. When E and γ were empirically measured via tensile tests, all five

materials exhibited compatible E, while FormLabs and ChronoSil showed a preferable lower γ.

Although the inspected strain region (ε = 10%) was not perfectly linear due to the nature of

materials’ nonlinear elasticity, our measurement results followed the general trend of previously

known material properties (Table 5). Leak testing was a critical qualitative elimination step that

ruled out three of the five material options. Only FormLabs Flexible and ChronoSil were qualified

for further assessments toward a functional DCCS. Volume amplification was limited by the small

size and cylindrical shape of our preliminary DCCS prototype (Figure 36A). Volume amplification

ratio is expected to increase with a modified sleeve design with a larger number of lumens formed

into a half-prolate shape for better coverage and fit around the curved epicardial surface of the

heart. As for energy requirements, the actuation energies required by both FormLabs and

116

ChronoSil prototypes fell within the range of generatable energy available from our implantable muscle-powered hydraulic power source (see Chapters 2 and 3), qualifying both as viable options for this particular application. Lap shear testing with biocompatible adhesives demonstrated that both FormLabs and ChronoSil provide sleeve-adhesive-epicardium complexes of sufficient strength to act as reliable fixation agents for DCCS attachment to the epicardium. During high frequency cyclical loading tests, the FormLabs prototype failed prematurely and was determined not suitable for long-term application. Again, despite the difficulty of defining a linear elastic region due to the elastomer’s nonlinearity, ChronoSil endured repetitive, high-speed deformations under rigorous conditions for 1 million cycles. The theoretical longevity of the ChronoSil prototype was also confirmed via FEA simulation using ANSYS Fatigue Module. FormLabs

Flexible resin is not biocompatible nor meant for medical implantation. However, ChronoSil is a corroborated biocompatible and biostable 3D printable polymer. Moveover, only ChronoSil passed cyclical loading tests and biocompatibility assessment for the ex vivo device implementation on fetal bovine hearts. It is also important to note that while LVEF can be expected to further increase with device design modifications, the possibility of cardiac compression using a ChronoSil DCCS was successfully demonstrated in this sub-optimal prototype. These results suggest that the ChronoSil material combined with modern 3D printing techniques that allow fabrication of biocompatible prototypes that are at the same time thin, flexible, and durable, can be used to thoroughly explore the design space for the refinement and production of soft robotic devices for direct cardiac compression.

In conclusion, these experiments, have established that ChronoSil is an acceptable material for 3D printing periventricular soft robotic actuators for the purposes of cardiac compression. The next step in the development process is to modify device design to safely and reliably achieve

117 clinically significant cardiac compressions for long-term circulatory support. In order to effectively magnify patients’ cardiac output without interfering with diastolic filling or coronary perfusion, the new sleeve should feature a larger size, increased number of lumens, and a curved wall profile. Its safety, efficacy, and pressure and force dynamics against a clinically pressurized heart could then be studied both ex and in vivo (Fig. 41).

Figure 41. Schematic of the bench setup for cVAD system assessment.

Each sleeve could also be customized on a patient-by-patient basis by rendering it around a 3D-scanned patient heart model using a CAD tool. The sleeve fit and function could then be simulated and optimized with a multiscale FEA simulations tool like Continuity Pro. This would allow a better understanding of temporal and spatial biomechanics, a more accurate stress-strain mapping of the epicardium, and insightful EF predictions valuable for effective and robust sleeve designs. Such a soft robotic cardiac compression system, once completed, would provide not only a safe, non-blood-contacting means to support the failing heart, but also offer a better patient quality-of-life over extended periods of time.

118

Chapter 6

Summary and Conclusions

The use of skeletal muscle as an endogenous power source is a promising approach to providing destination therapy for the tens of thousands of CHF patients who require long-term circulatory support each year. A truly fully-contained, tether-free system would offer significant clinical advantages over circulatory assist devices currently in use as well as significantly enhance patient quality of life. Moreover, due to its relative simplicity, it would likely be much less expensive to manufacture and maintain, resulting in wider availability and reduced health care costs. Toward that end, an implantable hydraulic muscle energy converter (MEC) powered by human latissimus dorsi muscle (LDM) was previously developed in this lab. In order to efficiently utilize and transmit the skeletal muscle energy captured by the MEC for the purposes of cardiac assistance, two different systems were developed in this thesis work.

The extra-aortic counterpulsation VAD (eVAD) was developed by combining the MEC with an extra-aortic balloon pump (EABP) using an implantable volume amplification mechanism

(iVAM) as an interface. Chapters 2 and 3 describe the iVAM design process, guided by the following seven design criteria: 1) anatomic fit, 2) muscle force and speed requirements, 3) work

119 storage and delivery, 4) energy transfer efficiency, 5) material selection, 6) volume amplification, and 7) device durability. Anatomic fit was achieved by engineering a linearly stacked MEC-iVAM complex designed to sit across the upper left ribcage and positioned directly beneath the humeral insertion of the LDM. The muscle force and actuation speed bench tests showed that the LDM is an appropriate power source for assisting hypertensive and tachycardiac patients. Results from muscle energy distribution analyses showed that roughly 94% of the contractile energy delivered to the MEC is either transmitted to the bloodstream or stored in the MEC-iVAM complex to power the return-strokes, qualifying the eVAD as an energy efficient system. Biocompatibility of the entire system was ensured by carefully selecting implant-safe, medical-grade materials. As for volume amplification, the iVAM amplified MEC output by 2.73 times, which fell short of the original volume amplification target of 4.0. However, refinements in manufacturing processes have been identified that are expected to eliminate this problem in next generation iVAM prototypes. The durability of the MEC and iVAM bellows and other internal components met the criteria for an “infinite” life rating. The longevity of the flexible EABP should be confirmed via cycle testing on the bench in future studies.

The cardiac compression copulsation VAD (cVAD) was developed by applying the MEC to a soft robotic direct cardiac compression sleeve (DCCS) that was designed and manufactured from scratch in this work. Chapter 4 describes the sleeve design optimization and biventricular model simulation processes. The preliminary sleeve design was optimized using computational modeling and static-structural finite element analysis. This was a pivotal pilot study that guided all subsequent sleeve design processes in terms of overall geometry, wall thickness, number of tubes, material hardness, and pressure resistance. The relationship, range, and effect of individual left ventricular (LV) and right ventricular (RV) epicardial compressions on a passive human

120 cardiac model was systematically examined next. A range of LV and RV epicardial pressures were applied against pressurized LV and RV computer models. The gap between the LV and RV epicardial compression requirements was found to be not only present but also rather large.

Although these results have limitations considering that these analyses were done on a simplified model and no patient heart will be completely passive prior to sleeve compression, it is safe to argue that separate applications of epicardial pressures on the LV and RV will be critical to maintaining comparable stroke volumes in both sides of the heart. This study is preliminary and incremental, but still provides fundamental insights and guidance towards the design of improved

DCC devices for long-term circulatory support.

Chapter 5 details the material and manufacturing method selection processes used for soft robotic cVAD development and describes an ex vivo demonstration of the device on a fetal bovine heart. Five 3D printable flexible polymers were tested as candidate sleeve body materials in terms of manufacturability, performance and functionality, and implantability. Tensile strength testing demonstrated that all five materials exhibit relatively compatible material characteristics.

Subsequent leak testing, however, ruled out three of the five material options, leaving only

FormLabs Flexible and ChronoSil qualified for further assessment. Volume amplification testing was limited by the small size and cylindrical shape of the preliminary prototype, but the possibility of fabricating a device of this type was demonstrated with both materials. Moreover, the actuation energies of both prototypes fell within the range of energy available from the MEC/LDM complex, indicating that the LDM is a viable power source for this application. Lap shear testing with biocompatible adhesives demonstrated that both Vetbond and GLUture provide sufficient adherence strength to act as reliable fixation agents for both FormLabs Flexible and ChronoSil sleeve materials. It was during high frequency cyclical loading tests when the FormLabs prototype

121 failed prematurely and was determined not suitable for long-term application. In contrast,

ChronoSil endured repetitive, high-speed deformations under rigorous conditions for 1 million cycles. Because ChronoSil is a proven biocompatible and biostable polymer, these results showed that of the five materials tested in this study only ChronoSil is suitable for 3D printing periventricular DCCS devices. Moreover, as an encouraging prelude to future studies, this initial

ChronoSil prototype induced a 55.6% left ventricular ejection fraction on a fetal bovine heart, thus demonstrating its potential as a long-term circulatory support mechanism.

The next step in the cVAD development process will be to modify device design for safe, reliable, and clinically significant cardiac output. In order to effectively magnify patients’ cardiac output without interfering with diastolic filling or coronary perfusion, future sleeve designs should feature a larger size, an increased number of lumens, and a curved wall profile. These new sleeves should also include a two-part compression system that supports the LV and RV independently from each other to optimize and balance left/right cardiac function as discussed in Chapter 4. To develop customized patient-specific devices, cardiac compressions and hemodynamics should be simulated using high-order biventricular models of patients’ beating hearts coupled with individualized fluid structure interaction analyses. This would provide a means to optimize the safety and efficacy of DCC therapy by providing a more accurate understanding of the compression requirements and cardiac behaviors unique to each patient. The newly optimized sleeve could then be prototyped by 3D printing using ChronoSil silicone elastomer.

In terms of immediate next steps and preclinical testing of the cVAD approach, the safety, efficacy, mechanical reliability, and pressure and force dynamics of next-generation prototype devices should first be studied against clinically pressurized hearts ex vivo. Acute (8 hour) and short-term (28 day) implant studies should then be performed in animals (i.e., pigs) with heart

122 failure to establish in vivo system functionality, document biological reactions at tissue/device interface surfaces, quantify prolonged hemodynamic effects, and determine the overall efficacy of the system as a whole. In subsequent clinical trials, because this unique muscle-powered VAD is, by definition, a Class III medical device with no analogous pre-approved predicates for comparison, the cVAD will be required to undergo an extensive, expensive and prolonged approval process prior to receiving clearance for widespread clinical use by the FDA. However, once approved this tether-free, non-blood-contacting VAD will be able to offer a safer means to support the failing heart while providing a better patient quality-of-life over extended periods of time.

123

References

1. Heart Failure | National Heart, Lung, and Blood Institute (NHLBI). https://www.nhlbi.nih.gov/health-topics/heart-failure.

2. Ambrosy, A. P. et al. The global health and economic burden of hospitalizations for heart failure: lessons learned from hospitalized heart failure registries. J. Am. Coll. Cardiol. 63, 1123–1133 (2014).

3. Mancini, D. & Colombo, P. C. Left Ventricular Assist Devices: A Rapidly Evolving Alternative to Transplant. J. Am. Coll. Cardiol. 65, 2542–2555 (2015).

4. Mandras, S. A., Crespo, J. & Patel, H. M. Innovative Application of Immunologic Principles in Heart Transplantation. Ochsner J 10, 231–235 (2010).

5. Lahpor, J. R. State of the art: implantable ventricular assist devices. Curr Opin Organ Transplant 14, 554–559 (2009).

6. Boyle, A. Current status of cardiac transplantation and mechanical circulatory support. Curr Heart Fail Rep 6, 28–33 (2009).

7. Birks, E. J. A Changing Trend Toward Destination Therapy. Tex Heart Inst J 38, 552–554 (2011).

8. Susen, S., Rauch, A., Van Belle, E., Vincentelli, A. & Lenting, P. J. Circulatory support devices: fundamental aspects and clinical management of bleeding and thrombosis. J Thromb Haemost 13, 1757–1767 (2015).

9. Xie, A., Phan, K. & Yan, T. D. Durability of continuous-flow left ventricular assist devices: a systematic review. Ann Cardiothorac Surg 3, 547–556 (2014).

10. Kilic, A., Emani, S., Sai-Sudhakar, C. B., Higgins, R. S. D. & Whitson, B. A. Donor selection in heart transplantation. J Thorac Dis 6, 1097–1104 (2014).

11. Holley, C. T., Harvey, L. & John, R. Left ventricular assist devices as a bridge to cardiac transplantation. J Thorac Dis 6, 1110–1119 (2014).

12. Melvin, D. B., Schima, H., Losert, U. M. & Wolner, E. Long-Term Ventricular Wall Actuation: Can and Should It Be Systematically Explored? Artificial Organs 20, 63–68 (1996).

13. Trumble, D. R., McGregor, W. E., Kerckhoffs, R. C. P. & Waldman, L. K. Cardiac assist with a twist: apical torsion as a means to improve failing heart function. J Biomech Eng 133, 101003 (2011).

124

14. Stewart, G. C. & Givertz, M. M. Mechanical circulatory support for advanced heart failure: patients and technology in evolution. Circulation 125, 1304–1315 (2012).

15. Anstadt, G. L., Schiff, P. & Baue, A. E. Prolonged circulatory support by direct mechanical ventricular assistance. Trans Am Soc Artif Intern Organs 12, 72–79 (1966).

16. Whitson, B. A. Surgical implant techniques of left ventricular assist devices: an overview of acute and durable devices. J Thorac Dis 7, 2097–2101 (2015).

17. Kilic, A. The future of left ventricular assist devices. J Thorac Dis 7, 2188–2193 (2015).

18. Prinzing, A. et al. Left ventricular assist devices—current state and perspectives. J Thorac Dis 8, E660–E666 (2016).

19. Cook, J. A. et al. The total artificial heart. J Thorac Dis 7, 2172–2180 (2015).

20. Englert, J. A. R., Davis, J. A. & Krim, S. R. Mechanical Circulatory Support for the Failing Heart: Continuous-Flow Left Ventricular Assist Devices. Ochsner J 16, 263–269 (2016).

21. Shah, S. P. & Mehra, M. R. Durable left ventricular assist device therapy in advanced heart failure: Patient selection and clinical outcomes. Indian Heart J 68, S45–S51 (2016).

22. Lee, L. S. et al. Ventricular restraint therapy for heart failure: the right ventricle is different from the left ventricle. J. Thorac. Cardiovasc. Surg. 139, 1012–1018 (2010).

23. Schmack, B. et al. Concurrent Left Ventricular Assist Device (LVAD) Implantation and Percutaneous Temporary RVAD Support via CardiacAssist Protek-Duo TandemHeart to Preempt Right Heart Failure. Med Sci Monit Basic Res 22, 53–57 (2016).

24. Chair, S. Y. et al. Evolvement of left ventricular assist device: the implications on heart failure management. J Geriatr Cardiol 13, 425–430 (2016).

25. Vetter, H. O. et al. Experience with the Novacor left ventricular assist system as a bridge to cardiac transplantation, including the new wearable system. The Journal of Thoracic and Cardiovascular Surgery 109, 74–80 (1995).

26. Rodriguez, L. E., Suarez, E. E., Loebe, M. & Bruckner, B. A. Ventricular Assist Devices (VAD) Therapy: New Technology, New Hope? Methodist Debakey Cardiovasc J 9, 32–37 (2013).

27. Gregory, S. D., Timms, D., Gaddum, N., Mason, D. G. & Fraser, J. F. Biventricular assist devices: a technical review. Ann Biomed Eng 39, 2313–2328 (2011).

28. Kiernan, M. S. et al. Early Right Ventricular Assist Device Use in Patients Undergoing Continuous-Flow Left Ventricular Assist Device Implantation: Incidence and Risk Factors

125

From the Interagency Registry for Mechanically Assisted Circulatory Support. Circ Heart Fail 10, (2017).

29. Haneya, A. et al. Temporary percutaneous right ventricular support using a centrifugal pump in patients with postoperative acute refractory right ventricular failure after left ventricular assist device implantation. Eur J Cardiothorac Surg 41, 219–223 (2012).

30. Rigatelli, G., Santini, F. & Faggian, G. Past and present of cardiocirculatory assist devices: a comprehensive critical review. J Geriatr Cardiol 9, 389–400 (2012).

31. Gerosa, G., Scuri, S., Iop, L. & Torregrossa, G. Present and future perspectives on total artificial hearts. Annals of Cardiothoracic Surgery 3, 595-602–602 (2014).

32. HeartMate II Left Ventricular Assist Device (LVAD) Fact Sheet.https://www.thoratec.com/downloads/heartmate-II-fact-sheet-b100-0812-final.pdf.

33. HeartMate IITMLVAD-Used in Over 25,000 Patients. https://www.heartmate.com/healthcare-provider/heartmate-ii-lvad.

34. Garbade, J., Bittner, H. B., Barten, M. J. & Mohr, F.-W. Current trends in implantable left ventricular assist devices. Cardiol Res Pract 2011, 290561 (2011).

35. Health, C. for D. and R. Recently-Approved Devices - HeartWareTM HVADTM - P100047/S090. https://www.fda.gov/MedicalDevices/ProductsandMedicalProcedures/DeviceApprovalsand Clearances/Recently-ApprovedDevices/ucm581473.htm.

36. HeartMate 3 Left Ventricular Assist System (LVAS)- P160054/S008.https://www.fda.gov/MedicalDevices/ProductsandMedicalProcedures/Devic eApprovalsandClearances/Recently-ApprovedDevices/ucm624155.htm

37. Thoratec CentriMag Blood Pump Fact Sheet.https://www.thoratec.com/downloads/CentriMag_Product_Fact_Sheet111109.pdf.

38. Timeline: First in VADS and Mechanical Assist Devices | Columbia University Department of Surgery. http://columbiasurgery.org/lvad/timeline-first-vads-and- mechanical-assist-devices.

39. Part II.9 – The Translational Pathway for Mechanical Circulatory Support – ISCTR | International Society for Cardiovascular Translational Research. https://isctr.org/chapter-ii- 9/.

40. Doty, D. Ventricular Assist Device and Destination Therapy Candidates from Preoperative Selection Through End of Hospitalization. Critical Care Nursing Clinics 27, 551–564 (2015).

126

41. Griffith, K., Jenkins, E., Stulak, J., Paugh, T. & Pagani, F. Long-term use of the CentriMag® Ventricular Assist System as a right ventricular assist device: a case report. Perfusion 27, 65–70 (2012).

42. Pediatric Mechanical Circulatory Support. Research Gate. https://www.researchgate.net/publication/272790183_Pediatric_Mechanical_Circulatory_S upport.

43. Heart Views. Official Journal of Gulf Heart Association.http://www.heartviews.org/viewimage.asp?img=HeartViews_2007_8_2_70_6 3742_f2.jpg.

44. Agnetti, G., Piepoli, M. F., Siniscalchi, G. & Nicolini, F. New Insights in the Diagnosis and Treatment of Heart Failure. Biomed Res Int 2015, 265260 (2015).

45. Park, W. H., Seo, Y. G. & Sung, J. D. Exercise Therapy for an Older Patient With Left Ventricular Assist Device. Ann Rehabil Med 38, 396–400 (2014).

46. HeartMate 3 Heart Pump Approved for Patients Not Eligible for Transplant |. Medgadget https://www.medgadget.com/2018/10/heartmate-3-heart-pump-approved-for-patients-not- eligible-for-transplant.html (2018).

47. Miller, L. W. & Guglin, M. Patient selection for ventricular assist devices: a moving target. J. Am. Coll. Cardiol. 61, 1209–1221 (2013).

48. Authority, U. of W. H. and C. Ventricular Assist Device (VAD) Patient Selection Criteria. UW Health https://www.uwhealth.org/heart-cardiovascular/ventricular-assist-device-vad- patient-selection-criteria/39748.

49. Trinsey, A. Extracorporeal membrane oxygenation: A review. Nursing2018 Critical Care 12, 16 (2017).

50. Extracorporeal Membrane Oxygenation (ECMO) | Frequently Asked Questions | Boston Children’s Hospital. http://www.childrenshospital.org/centers-and-services/programs/a-_- e/extracoporeal-membrane-oxygenation-ecmo-program/frequently-asked-questions-for- parents.

51. Health, C. for D. and R. CDRH Transparency - Reclassification. https://www.fda.gov/AboutFDA/CentersOffices/OfficeofMedicalProductsandTobacco/CD RH/CDRHTransparency/ucm378724.htm.

52. Mosier, J. M. et al. Extracorporeal membrane oxygenation (ECMO) for critically ill adults in the emergency department: history, current applications, and future directions. Crit Care 19, 431 (2015).

127

53. Rossiter-Thornton, M., Arun, V., Forrest, A. P., Bayfield, M. S. & Wilson, M. K. Left Ventricular Support with the Impella® LP 5.0 for Cardiogenic Shock Following Cardiac Surgery. Heart, Lung and Circulation 17, 243–245 (2008).

54. Naidu, S. S. Novel Percutaneous Cardiac Assist Devices: The Science of and Indications for Hemodynamic Support. Circulation 123, 533–543 (2011).

55. Abiomed Receives Approval for Expanded FDA Indication for Cardiomyopathy with Cardiogenic Shock. ABIOMED, Inc. http://investors.abiomed.com/news-releases/news- release-details/abiomed-receives-approval-expanded-fda-indication-cardiomyopathy.

56. Abiomed Receives Approval for Expanded FDA Indication for High Risk Percutaneous Coronary Intervention (PCI) Procedures. ABIOMED, Inc. http://investors.abiomed.com/news-releases/news-release-details/abiomed-receives- approval-expanded-fda-indication-high-risk-0.

57. Health, C. for D. and R. Recently-Approved Devices - Impella Ventricular Support Systems - P140003/S018. https://www.fda.gov/medicaldevices/productsandmedicalprocedures/deviceapprovalsandcl earances/recently-approveddevices/ucm598801.htm.

58. Impella Ventricular Support Systems for Use During Cardiogenic Shock and High-Risk PIC. ABIOMED.154202963952329325f54536fc43bdfb5343fbf029.pdf

59. Parissis, H. et al. IABP: history-evolution-pathophysiology-indications: what we need to know. J Cardiothorac Surg 11, (2016).

60. 510(k) Premarket Notification. https://www.accessdata.fda.gov/SCRIPTs/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K790241.

61. Counterpulsation Applied. An Introduction to Intra-Aortic Balloon Pumping. TELEFLEX. MC-000363_Counterpulsation-Manual.pdf.

62. Thoratec Corporate Fact Sheet.https://www.thoratec.com/downloads/GL-THO-03150196- Thoratec%20Corporate%20Fact%20Sheet-FINAL.pdf

63. 2ndGeneration CentriMag Primary Console Operating Manual. Thoratec. PL- 0047_Rev_08_2nd_Generation_CentriMag_System_Operating_Manual_(US).pdf.

64. ECMO. The Dept of Anaesthesia andIntensive Care.https://www.aic.cuhk.edu.hk/web8/ecmo.htm.

65. Samuels, L. E., Casanova-Ghosh, E. & Droogan, C. Cardiogenic shock associated with loco-regional anesthesia rescued with left ventricular assist device implantation. J Cardiothorac Surg 5, 126 (2010).

128

66. Intra-aortic Balloon Pump (IABP) Market 2017 - Maquet (Getinge Group), Teleflex Incorporated - openPR. https://www.openpr.com/news/706293/Intra-aortic-Balloon-Pump- IABP-Market-2017-Maquet-Getinge-Group-Teleflex-Incorporated.html.

67. Kazui, T. et al. Minimally invasive approach for percutaneous CentriMag right ventricular assist device support using a single PROTEKDuo Cannula. J Cardiothorac Surg 11, 123 (2016).

68. HeartMate 3 Left Ventricular Assist System Instructions for Use. Thoratec Coporation.https://www.accessdata.fda.gov/cdrh_docs/pdf16/P160054C.pdf.

69. Mehra, M. R. et al. Two-Year Outcomes with a Magnetically Levitated Cardiac Pump in Heart Failure. N. Engl. J. Med. 378, 1386–1395 (2018).

70. MOMENTUM 3 data shows improved survival for HeartMate 3 heart pump. Cardiac Rhythm News https://cardiacrhythmnews.com/new-long-term-data-show-improved- survival-and-lower-rates-of-stroke-and-pump-thrombosis-for-abbotts-heartmate-3-heart- pump/ (2018).

71. HeartWare Ventricular Assist System Instructions for Use.http://www.heartware.com/sites/default/files/uploads/docs/ifu00001_rev_15.pdf.

72. Villa, C. R. & Morales, D. L. S. The Total Artificial Heart in End-Stage Congenital Heart Disease. Front. Physiol. 8, (2017).

73. Total Artificial Heart Etiologies & Outcomes. SynCardia https://syncardia.com/clinicians/clinical-resources/patients-outcomes-experience/ (2018).

74. Bartfay, S. et al. (675) - Berlin Heart Excor as Bridge to Heart Transplantation. A Versatile Option in Patients Not Suitable for a Continuous Flow LVAD. The Journal of Heart and Lung Transplantation 37, S271 (2018).

75. EXCOR Pediatric VAD Instructions for Use.https://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/P ediatricAdvisoryCommittee/UCM537702.pdf

76. Bockeria, L. A. et al. Miniature Rotary Blood Pumps for Use in Pediatric Cardiac Surgery. Biomed Eng 50, 291–295 (2017).

77. Lima, B. et al. Controversies and Challenges of Ventricular Assist Device Therapy. Am. J. Cardiol. 121, 1219–1224 (2018).

78. Kirklin, J. K. et al. Eighth annual INTERMACS report: Special focus on framing the impact of adverse events. J. Heart Lung Transplant. 36, 1080–1086 (2017).

129

79. Nienaber, J. J. C. et al. Clinical manifestations and management of left ventricular assist device-associated infections. Clin. Infect. Dis. 57, 1438–1448 (2013).

80. Leuck, A.-M. Left ventricular assist device driveline infections: recent advances and future goals. J Thorac Dis 7, 2151–2157 (2015).

81. Harvey, L., Holley, C. T. & John, R. Gastrointestinal bleed after left ventricular assist device implantation: incidence, management, and prevention. Ann Cardiothorac Surg 3, 475–479 (2014).

82. Cushing, K. & Kushnir, V. Gastrointestinal Bleeding Following LVAD Placement from Top to Bottom. Dig Dis Sci 61, 1440–1447 (2016).

83. Ventricular Assist Device Exit Site Care - ppt video online download. https://slideplayer.com/slide/1442089/.

84. Uriel, N. et al. Development of a novel echocardiography ramp test for speed optimization and diagnosis of device thrombosis in continuous-flow left ventricular assist devices: the Columbia ramp study. J. Am. Coll. Cardiol. 60, 1764–1775 (2012).

85. Lünse, S., Höhn, J., Simon, P., Heidecke, C.-D. & Glitsch, A. Dieulafoy-Like Lesion at the Brim of a Gastric Diverticulum: A Very Rare Cause of Gastrointestinal Bleeding. Journal of Gastrointestinal & Digestive System 8, 1–1 (2018).

86. Magovern, G. J., Park, S. B., Magovern, G. J., Christlieb, I. Y. & Kao, R. L. Mechanical Circulatory Assist Devices. Tex Heart Inst J 14, 276–283 (1987).

87. Tchantchaleishvili, V., Bush, B. S., Swartz, M. F., Day, S. W. & Massey, H. T. Plutonium- 238: an ideal power source for intracorporeal ventricular assist devices? ASAIO J. 58, 550– 553 (2012).

88. LaForge, D. & Portner, P. An efficient pulsed-solenoid powered cardiac assist system. IEEE Transactions on Magnetics 18, 1481–1483 (1982).

89. Farrar, D. J., Buck, K. E., Coulter, J. H. & Kupa, E. J. Portable pneumatic biventricular driver for the Thoratec ventricular assist device. ASAIO J. 43, M631-634 (1997).

90. Najjar, E. et al. Controller and battery changes due to technical problems related to the HVAD® left ventricular assist device - a single center experience. J Cardiothorac Surg 13, (2018).

91. Dunlay, S. M. et al. Dying with a Left Ventricular Assist Device as Destination Therapy. Circ Heart Fail 9, (2016).

130

92. Bocan, K. N. & Sejdić, E. Adaptive Transcutaneous Power Transfer to Implantable Devices: A State of the Art Review. Sensors (Basel) 16, (2016).

93. Wang, J. X., Smith, J. R. & Bonde, P. Energy Transmission and Power Sources for Mechanical Circulatory Support Devices to Achieve Total Implantability. The Annals of Thoracic Surgery 97, 1467–1474 (2014).

94. Basaeri, H., Christensen, D. B. & Roundy, S. A review of acoustic power transfer for bio- medical implants. Smart Mater. Struct. 25, 123001 (2016).

95. Denisov, A. & Yeatman, E. M. Ultrasonic vs. Inductive Power Delivery for Miniature Biomedical Implants. 2010 International Conference on Body Sensor Networks 84–89 (2010) doi:10.1109/BSN.2010.27.

96. Trumble, D. R. & Webster, J. G. Artificial Hearts and Cardiac Assist Devices: Performance, Control, and Power Delivery. in Wiley Encyclopedia of Electrical and Electronics Engineering (John Wiley & Sons, Inc., 1999). doi:10.1002/047134608X.W6602.pub2.

97. What are the limitations of the AbioCor TAH for the treatment of heart failure? https://www.medscape.com/answers/163062-86476/what-are-the-limitations-of-the- abiocor-tah-for-the-treatment-of-heart-failure.

98. Lionheart recipient at Penn State Hershey Medical Center first US patient to head home with device. EurekAlert! http://www.eurekalert.org/pub_releases/2003-06/ps- lra062403.php.

99. Puers, R. & Vandevoorde, G. Recent progress on transcutaneous energy transfer for total artificial heart systems. Artif Organs 25, 400–405 (2001).

100. Enssle, A. Transcutaneous Energy Transfer for Left Ventricular Assist Devices :: Create the Future Design Contest. http://contest.techbriefs.com/2016/entries/medical/7069.

101. Dissanayake, T., Budgett, D., Hu, A. P., Malpas, S. & Bennet, L. Transcutaneous Energy Transfer System for Powering Implantable Biomedical Devices. in 13th International Conference on Biomedical Engineering (eds. Lim, C. T. & Goh, J. C. H.) 235–239 (Springer Berlin Heidelberg, 2009).

102. Trumble, D. R. & Magovern, J. A. A muscle-powered energy delivery system and means for chronic in vivo testing. J. Appl. Physiol. 86, 2106–2114 (1999).

103. Trumble, D. R. A Muscle-Powered Counterpulsation Device for Tether-Free Cardiac Support: Form and Function 1. Journal of Medical Devices 10, 020903 (2016).

131

104. Trumble, D. & A Magovern, J. Linear Muscle Power for Cardiac Support: a Progress Report. Basic and applied myology: BAM 9, (1999).

105. Salmons, S. & Jarvis, J. The working capacity of skeletal muscle transformed for use in a cardiac assist role. Futura Publ Co 89–104 (1990).

106. Gray, Henry. 1918. Anatomy of the Human Body. https://www.bartleby.com/107/.

107. Han, J., Kubala, M. & Trumble, D. R. Design of a Muscle-Powered Soft Robotic Bi-VAD for Long-Term Circulatory Support. V001T01A003 (2018) doi:10.1115/DMD2018-6835.

108. Mitsumaru, A. et al. Experimental study of combination of extraaortic balloon counterpulsation and ventricular assist cup to acute heart failure in dogs. ASAIO J. 43, 187– 192 (1997).

109. Lowe, J. E. et al. First successful bridge to cardiac transplantation using direct mechanical ventricular actuation. Ann. Thorac. Surg. 52, 1237–1243; discussion 1243-1245 (1991).

110. Oz, M. C., Artrip, J. H. & Burkhoff, D. Direct cardiac compression devices. The Journal of Heart and Lung Transplantation 21, 1049–1055 (2002).

111. Conte, J. V. Surgical Ventricular Restoration: Technique and Outcomes. Congestive Heart Failure 10, 248–251 (2004).

112. BBC NEWS | Health | Artificial heart muscle tested. http://news.bbc.co.uk/2/hi/health/3151891.stm.

113. Roche, E. T. et al. Soft robotic sleeve supports heart function. Science Translational Medicine 9, eaaf3925 (2017).

114. corinnova | Preclinical Results. corinnova https://www.corinnova.com/preclin.

115. Reesink, K. D. et al. Miniature intracardiac assist device provides more effective cardiac unloading and circulatory support during severe left heart failure than intraaortic balloon pumping. Chest 126, 896–902 (2004).

116. C-Pulse® System: A Heart Assist Device Clinical Study - Full Text View - ClinicalTrials.gov. https://clinicaltrials.gov/ct2/show/NCT01740596.

117. C-Pulse Extra-Aortic Heart Pump Wins Approvals in Europe |. Medgadget https://www.medgadget.com/2012/07/c-pulse-extra-aortic-heart-pump-wins-approvals-in- europe.html (2012).

118. Kwon, M. H., Cevasco, M., Schmitto, J. D. & Chen, F. Y. Ventricular restraint therapy for heart failure: A review, summary of state of the art, and future directions. J. Thorac. Cardiovasc. Surg. 144, 771-777.e1 (2012).

132

119. Klodell, C. T. et al. Worldwide surgical experience with the Paracor HeartNet cardiac restraint device. J. Thorac. Cardiovasc. Surg. 135, 188–195 (2008).

120. Health, C. for D. and R. HeartWareTM HVADTM - P100047/S090. FDA (2019).

121. Long, J. W. et al. Long-term destination therapy with the HeartMate XVE left ventricular assist device: improved outcomes since the REMATCH study. Congest Heart Fail 11, 133–138 (2005).

122. Jaquiss, R. D. B. et al. Postapproval Outcomes: The Berlin Heart EXCOR Pediatric in North America. ASAIO Journal 63, 193 (2017).

123. Pennington, D. G. & Lohmann, D. P. Novacor LVAS Implantation Technique. Operative Techniques in Thoracic and Cardiovascular Surgery 4, 318–329 (1999).

124. PVAD IVAD Fact Sheet-B100-0412-FINAL.pdf.

125. Lad, V. et al. Mechanical circulatory support with the ABIOMED BVS 5000: The Toronto General Hospital experience. Can J Cardiol 26, 467–470 (2010).

126. Levinson, M. M. et al. Thromboembolic complications of the Jarvik-7 total artificial heart: case report. Artif Organs 10, 236–244 (1986).

127. Robert Jarvik, MD on the Jarvik-7. Jarvik Heart Inc. https://www.jarvikheart.com/history/robert-jarvik-on-the-jarvik-7/ (2016).

128. Gilotra, N. A. & Stevens, G. R. Temporary Mechanical Circulatory Support: A Review of the Options, Indications, and Outcomes. Clin Med Insights Cardiol 8, 75–85 (2015).

129. Class 2 Device Recall Jarvik 2000 Ventricular Assist System. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfRes/res.cfm?ID=169598.

130. Schroder, J. N. & Milano, C. A. A tale of two centrifugal left ventricular assist devices. The Journal of Thoracic and Cardiovascular Surgery 154, 850–852 (2017).

131. Health, C. for D. and R. Recently-Approved Devices - HeartMate 3 Left Ventricular Assist System (LVAS) - P160054/S008. https://www.fda.gov/MedicalDevices/ProductsandMedicalProcedures/DeviceApprovalsand Clearances/Recently-ApprovedDevices/ucm624155.htm.

132. Katz, J. N., Waters, S. B., Hollis, I. B. & Chang, P. P. Advanced Therapies For End-Stage Heart Failure. Curr Cardiol Rev 11, 63–72 (2015).

133. Melvin, D. B. Device-Induced Ventricular Geometric Remodeling: Appraisal of Critical Issues. Journal of Cardiac Surgery 16, 34–39 (2001).

133

134. Milano, C. A. et al. HVAD: The ENDURANCE Supplemental Trial. JACC: Heart Failure 6, 792–802 (2018).

135. Care, I. of M. (US) and N. A. of E. (US) R. on V. & S.-D. H. Healthcare System Complexities, Impediments, and Failures. (National Academies Press (US), 2011).

136. Trumble, D. R. & Magovern, J. A. Muscle-Powered Mechanical Blood Pumps. Science 296, 1967–1967 (2002).

137. Aranda-Michel, E., Han, J. & Trumble, D. R. Design of a Muscle-Powered Extra-Aortic Counterpulsation Device for Long-Term Circulatory Support. in 2017 Design of Medical Devices Conference (American Society of Mechanical Engineers, 2017). doi:10.1115/DMD2017-3325.

138. Trumble, D. R., Norris, M. & Melvin, A. Design Improvements and In Vitro Testing of an Implantable Muscle Energy Converter for Powering Pulsatile Cardiac Assist Devices. J. Med. devices 4, (2010).

139. Trumble, D. R., Melvin, D. B., Byrne, M. T. & Magovern, J. A. Improved Mechanism for Capturing Muscle Power for Circulatory Support. Artif Organs 29, 691–700 (2005).

140. Kantrowitz, A. et al. Initial clinical experience with a new permanent mechanical auxiliary ventricle: the dynamic aortic patch. Trans Am Soc Artif Intern Organs 18, 159–167, 179 (1972).

141. Kantrowitz, A. Experimental augmentation of coronary flow by retardation of the arterial pressure pulse. Surgery 34, 678–687 (1953).

142. Kantrowitz, A. Origins of intraaortic balloon pumping. Ann. Thorac. Surg. 50, 672–674 (1990).

143. Legget, M. E. et al. Extra-Aortic Balloon Counterpulsation. Circulation 112, I-26-I–31 (2005).

144. Mitnovetski, S. et al. Extra-aortic implantable counterpulsation pump in chronic heart failure. Ann. Thorac. Surg. 85, 2122–2125 (2008).

145. Izzo, J., Gabbay, S. & Mayott, C. In vitro testing and evaluation of intraaortic and extraaortic balloon counterpulsation devices. in Proceedings of 1994 20th Annual Northeast Bioengineering Conference 124- (1994). doi:10.1109/NEBC.1994.305154.

146. VENTURELLI, C., JEANNIN, G., SOTTINI, L., DALLERA, N. & SCOLARI, F. Cholesterol crystal embolism (atheroembolism). Heart Int 2, (2006).

134

147. Rafieian-Kopaei, M., Setorki, M., Doudi, M., Baradaran, A. & Nasri, H. Atherosclerosis: Process, Indicators, Risk Factors and New Hopes. Int J Prev Med 5, 927–946 (2014).

148. Schulz, A. et al. Preliminary Results From the C-Pulse OPTIONS HF European Multicenter Post-Market Study. Med Sci Monit Basic Res 22, 14–19 (2016).

149. Sales, V. L. & McCarthy, P. M. Understanding the C-pulse device and its potential to treat heart failure. Curr Heart Fail Rep 7, 27–34 (2010).

150. de Waha, S. et al. Intra-aortic balloon counterpulsation — Basic principles and clinical evidence. Vascular Pharmacology 60, 52–56 (2014).

151. Trumble, D. Linear muscle power for cardiac support: Current progress and future directions. Basic Applied Myology 19, 35–40 (2009).

152. Gopinathannair, R. & Olshansky, B. Management of tachycardia. F1000Prime Rep 7, (2015).

153. Bibas, L., Levi, M. & Essebag, V. Diagnosis and management of supraventricular tachycardias. CMAJ 188, E466–E473 (2016).

154. AM 350 Tech Data. https://www.hightempmetals.com/techdata/hitempAM350data.php.

155. AM-350-Stainless-Steel-Wire-UNS-S35000.pdf.

156. Lloyd-Jones, D. M. et al. Lifetime risk for developing congestive heart failure: the Framingham Heart Study. Circulation 106, 3068–3072 (2002).

157. Mahmud, A., Hennessy, M. & Feely, J. Effect of sildenafil on blood pressure and arterial wave reflection in treated hypertensive men. J Hum Hypertens 15, 707–713 (2001).

158. He´bert, J.-L. et al. Relation between aortic dicrotic notch pressure and mean aortic pressure in adults. American Journal of Cardiology 76, 301–306 (1995).

159. Kamada, T. et al. Antihypertensive efficacy and safety of the angiotensin receptor blocker azilsartan in elderly patients with hypertension. Drug Chem Toxicol 40, 110–114 (2017).

160. Saliba, E. & Sia, Y. The ascending aortic aneurysm: When to intervene? Int J Cardiol Heart Vasc 6, 91–100 (2015).

161. Hong, M. S. et al. Ectatic Aortas (2.5-2.9 cm) Are At Risk For Progression To Abdominal Aortic Aneurysm. Annals of Vascular Surgery 34, 27–28 (2016).

162. Elefteriades, J. A. & Farkas, E. A. Thoracic aortic aneurysm clinically pertinent controversies and uncertainties. J. Am. Coll. Cardiol. 55, 841–857 (2010).

135

163. Trumble, D. R. & Magovern, J. A. A permanent prosthesis for converting in situ muscle contractions into hydraulic power for cardiac assist. J. Appl. Physiol. 82, 1704–1711 (1997).

164. Ventricle-specific epicardial pressures as a means to optimize direct cardiac compression for circulatory support: A pilot study. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0219162.

165. Han, J. & Trumble, D. R. Cardiac Assist Devices: Early Concepts, Current Technologies, and Future Innovations. Bioengineering 6, 18 (2019).

166. Mann, D. L. et al. Clinical evaluation of the CorCap Cardiac Support Device in patients with dilated cardiomyopathy. Ann. Thorac. Surg. 84, 1226–1235 (2007).

167. Shanmugam, G. & Ali, I. S. Surgical Ventricular Restoration: An Operation to Reverse Remodeling - Clinical Application (Part II). Curr Cardiol Rev 5, 350–359 (2009).

168. Heart Failure: Types, Symptoms, Causes & Treatments. https://my.clevelandclinic.org/health/diseases/17069-heart-failure-understanding-heart- failure.

169. Kung, R. T. V. & Rosenberg, M. Heart Booster: a pericardial support device. The Annals of Thoracic Surgery 68, 764–767 (1999).

170. Eidem, B. W., Cetta, F. & O’Leary, P. W. Echocardiography in Pediatric and Adult Congenital Heart Disease. (Lippincott Williams & Wilkins, 2012).

171. Cwajg, J. M. et al. End-diastolic wall thickness as a predictor of recovery of function in myocardial hibernation: Relation to rest-redistribution Tl-201 tomography and dobutamine stress echocardiography. Journal of the American College of Cardiology 35, 1152–1161 (2000).

172. Ascione, L. et al. End-Diastolic Wall Thickness as a Predictor of Reverse Remodelling After Cardiac Resynchronization Therapy: A Two-Dimensional Echocardiographic Study. Journal of the American Society of Echocardiography 21, 1055–1061 (2008).

173. Wagholikar, K. B. et al. Extraction of Ejection Fraction from Echocardiography Notes for Constructing a Cohort of Patients having Heart Failure with reduced Ejection Fraction (HFrEF). J Med Syst 42, (2018).

174. El Missiri, A. M., El Meniawy, K. A. L., Sakr, S. A. S. & Mohamed, A. S. E. Normal reference values of echocardiographic measurements in young Egyptian adults. The Egyptian Heart Journal 68, 209–215 (2016).

136

175. Fuchs, A. et al. Normal values of left ventricular mass and cardiac chamber volumes assessed by 320-detector computed tomography angiography in the Copenhagen General Population Study. Eur Heart J Cardiovasc Imaging 17, 1009–1017 (2016).

176. Kovalova, S., Necas, J. & Vespalec, J. What is a “normal” right ventricle? Eur J Echocardiogr 7, 293–297 (2006).

177. Hagan Arthur D. et al. Ultrasound Evaluation of Systolic Anterior Septal Motion in Patients With and Without Right Ventricular Volume Overload. Circulation 50, 248–254 (1974).

178. Soohoo, E., Waldman, L. K. & Trumble, D. R. Computational Parametric Studies Investigating the Global Hemodynamic Effects of Applied Apical Torsion for Cardiac Assist. Ann Biomed Eng 45, 1434–1448 (2017).

179. Guccione, J. M., McCulloch, A. D. & Waldman, L. K. Passive Material Properties of Intact Ventricular Myocardium Determined From a Cylindrical Model. J Biomech Eng 113, 42– 55 (1991).

180. Armstrong, D. W., Tsimiklis, G. & Matangi, M. F. Factors influencing the echocardiographic estimate of right ventricular systolic pressure in normal patients and clinically relevant ranges according to age. Can J Cardiol 26, e35–e39 (2010).

181. Redington, A. N., Gray, H. H., Hodson, M. E., Rigby, M. L. & Oldershaw, P. J. Characterisation of the normal right ventricular pressure-volume relation by biplane angiography and simultaneous micromanometer pressure measurements. Br Heart J 59, 23–30 (1988).

182. CV Physiology | Ventricular Pressure-Volume Relationship. https://www.cvphysiology.com/Cardiac%20Function/CF024.

183. Heart Failure Classifications | NYHA Classification. CHF Solutions https://www.chf- solutions.com/heart-failure-classifications/.

184. 3D printing of soft robotic systems | Nature Reviews Materials. https://www.nature.com/articles/s41578-018-0002-2.

185. 3D printing for soft robotics – a review: Science and Technology of Advanced Materials: Vol 19, No 1. https://www.tandfonline.com/doi/full/10.1080/14686996.2018.1431862.

186. Expert-guided optimization for 3D printing of soft and liquid materials. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0194890.

187. 3D printing of robotic soft actuators with programmable bioinspired architectures | Nature Communications. https://www.nature.com/articles/s41467-018-03216-w.

137

188. FormLabs Flexible Resin Material Data Sheet. https://formlabs- media.formlabs.com/datasheets/Flexible_Technical.pdf

189. PolyFlexTM by Polymaker Tech sheet | 3D FilaPrint. http://3dfilaprint.com/polyflex-by- polymaker-tech-sheet/.

190. FILAFLEX 82A TECHNICAL SHEET & MSDS 2019.pdf. https://app.box.com/s/daro2m4t5mskfrdoeie01m9o5y718x5y

191. NinjaTekTM NinjaFlex Technical Data Sheet 2016 | 3D FilaPrint. http://3dfilaprint.com/ninjatek-ninjaflex-technical-data-sheet-2016/.

192. ChronoSil. http://www.advbiomaterials.com/products/polycarbonate/chronosil.html.

193. ASTM D638 - 14 Standard Test Method for Tensile Properties of Plastics. https://www.astm.org/Standards/D638.

194. Muscle-powered Counterpulsation for Untethered, Non-blood-contacting Cardiac Support: A Path to Destination Therapy - IEEE Journals & Magazine. https://ieeexplore.ieee.org/abstract/document/9001177.

195. ASTM D1002 - 10(2019) Standard Test Method for Apparent Shear Strength of Single- Lap-Joint Adhesively Bonded Metal Specimens by Tension Loading (Metal-to-Metal). https://www.astm.org/Standards/D1002.htm.

196. 3M Vetbond Tissue Adhesive 1469 Safety Data Sheet. 3m-vetbond.pdf.

197. GLUture Topical Tissue Adhesive Prodict Information. https://www.zoetisus.com/contact/pages/product_information/msds_pi/pi/gluture.pdf

198. Development of Epicardial Circulatory Assist Devices: Material Considerations | Frontiers in Biomedical Devices | ASME Digital Collection. https://asmedigitalcollection.asme.org/BIOMED/proceedings/DMD2018/40789/V001T01 A010/271881.

199. ASTM D5026 Tension Dynamic Plastics Composites Test Equipment. https://www.testresources.net/applications/standards/astm/astm-d5026-tension-dynamic- plastics-composites-test-equipment/.

200. Are Formlabs resins food-safe or biocompatible? https://support.formlabs.com/s/article/Are-Formlabs-resins-food-safe-or- biocompatible?language=en_US.

201. Tough adhesives for diverse wet surfaces | Science. https://science.sciencemag.org/content/357/6349/378.

138

202. Frontiers | Therapeutic adhesives: covalently modified acrylates for enhanced, injury- specific wound healing. https://www.frontiersin.org/10.3389/conf.FBIOE.2016.01.00129/2893/10th_World_Biomat erials_Congress/all_events/event_abstract.

139